Você está na página 1de 246

CARY WOLFE, S E R I E S EDITOR

15

Animal Stories: Narrating across Species Lines

ISABELLE STENGERS
TRANSLATED

BY

ROBERT

BONONNO

Susan McHugh
14

Human Error: Species -Being and Media Machines


Dominic Pettman

13

Junkware
Thierry Bardini

12

A Foray into the Worlds ofAnimals and Humans.


with A Theory of Meaning

Cosmopolitics II

Jakob von Uexkull


11

Insect Media: An Archaeology of Animals and Technology


Jussi Parikka

10

Cosmopolitics II
Isabelle Stengers

Cosmopolitics I
Isabelle Stengers

IV. Quantum Mechanics


V. In the Name of the Arrow of Time
VI. Life and Artifice
VII. The Curse of Tolerance

What Is Posthumanism?
Caiy Wolfe

Political Affect: Connecting the Social and the Somatic

p o s t h u m a n i t i e s 10

John Protevi
6

Animal Capital: Rendering Life in Biopolitical Times


Nicole Shukin

Dorsality: Thinking Back through Technology and Politics


David Wills

4,

Bios: Biopolitics and Philosophy


Roberto Esposito

When Species Meet


Donna J. Haraway

The Poetics ofDNA


Judith Roof

The Parasite
Michel Serres

IN
NE
SO
TA
U N I V E R S I T Y OF M I N N E S O T A
MINNEAPOLIS

LONDON

PRESS

The University of Minnesota Press gratefully acknowledges the generous


assistance provided for the publication of this book by the Hamilton P. Traub

CONTENTS

University Press Fund.


Published in French as Cosmopolitiques II; copyright Editions La Decouverte,
2003. The contents of this book originally appeared as part of a seven-volume
edition of Cosmopolitiques. also published by Editions La Decouverte in 1997.
Cosmopolitiques I includes volumes 1. II, and III of the original edition, and
Cosmopolitiques II includes volumes IV. V. VI, and VII of the original edition.
English translation copyright 2011 by Robert Bononno

Preface /vii

All rights reserved. No part of this publication may be reproduced, stored in

BOOK I V . QUANTUM M E C H A N I C S : T H E END OF T H E D R E A M

a retrieval system, or transmitted, in any form or by any means, electronic,

1. Atoms Exist! / j

mechanical, photocopying, recording, or otherwise, without the prior written

2. Abandon the Dream? / 1 8

permission of the publisher.

3. Niels Bohr's Lesson

Published by the University of Minnesota Press


111 Third Avenue South, Suite 290

4. Quantum Irony /48


5. The Physicists' Double Standard / 71
6. The Silent Descendant of the Queen of Heaven / 86

Minneapolis, MN 55401-2520
http://www.upres8.umn.edu
Library of Congress Cataloging-in-Publication Data

BOOK V. I N T H E N A M E OF T H E ARROW OF T I M E :
PRIGOGINE'S CHALLENGE

Stengers, Isabelle.

8. Boltzmann's Successor /123

7. The Arrow of Time / 1 0 5


[Cosmopolitiques. English]

9.

Cosmopolitics / Isabelle Stengers; translated by Robert Bononno.


v. cm. (Posthumanities ; 10-)

Boltzmann's Heir/14,3

10. The Obligations of Chaos/160


11. The Laws of Chaos? /176

Includes bibliographical references and index.


Contents: 1. Quantum Mechanics, In the Name of the Arrow of Time, Life and
Artifice, The Curse of Tolerance
ISBN 978-0-8166-5688-2 (v. 2 : he : alk. paper) - ISBN 978-0-8166-5689-9
(v. 2 : pb: alk. paper)
1. ScienceHistory. 2. SciencePhilosophy. 3. ScienceSocial aspects. I. Title.
Q125.S74.2613 2010

12. The Passion of the Law / 292


BOOK V I . L I F E A N D A R T I F I C E :
T H E F A C E S OF E M E R G E N C E
13. The Question of Emergence / 207
14. The Practices of Emergence / 319
15. Dissipative Coherence /234

501dc22

16. Artifice and Life / 250

2010010387

17. The Art of Models / 269


18. Transition to the Limit/2S4
Printed in the United States of America on acid-free paper
The University of Minnesota is an equal-opportunity educator and employer.
18 17 16 15 14 13 12 11

10

987654321

PREFACE

BOOK V I I . T H E C U R S E OF T O L E R A N C E

How can we examine the discordant landscape of knowledge

19. The Curse of Tolerance / 303

derived from modern science? Is there any consistency to be

20. The Curse as Test/312

found among contradictoiy or mutually exclusive visions, ambi-

2,\. Anxiety and Fright/324

tions, and methods? Is the hope of a "new alliance" that was

22. The Politics of Technical Inventions /340


23. The Cosmopolitical Question /35.2
24. Nomadic and Sedentary / 363
25. The Betrayal of the Diplomats /374

expressed more than twenty years ago destined to remain a hollow dream?
I would like to respond to these questions by arguing for

26. The Diplomats' Peace /386

an "ecology of practices." I have constructed my argument in

27. Calculemus / 393

seven steps or parts, covering two separate volumes1this is the

28. The Final Challenge /407

second. Each section is self-contained and can be read on its

Notes / 417
Index / 457

own, but I would hope that readers view individual sections as


an invitation to read the others, for the collection forms a unified whole. Step by step, I attempt to bring into existence seven
problematic landscapes, seven endeavors to create the possibility of consistency where there is currently only confrontation.
Whether the topic is the nature of physics and physical law, the
debate over self-organization and emergence, or the challenges
posed by ethnopsychiatiy to the division between modern and
archaic knowledge, in each case I tried to address the practices
from which such knowledge evolves, based on the constraints
imposed by the uncertainties they introduce and their corresponding obligations. No unifying body of knowledge will
ever demonstrate that the neutrino of physics can coexist with
the multiple worlds mobilized by ethnopsychiatiy. Nonetheless, such coexistence has a meaning, and it has nothing to do
with tolerance or disenchanted skepticism. Such beings can be

vii

viii

PREFACE

collectively affirmed in a "cosmopolitical" space in which collide the hopes and doubts and fears and dreams they engender

BOOK IV

and which cause them to exist. That is why, through the exploration of knowledge, what I would like to convey to the reader is
also a form of ethical experimentation.

Quantum Mechanics
THE

END

OF T H E

DREAM

Atoms Exist!

In the early twentieth century a page was turned in the history


of physics, which was now dedicated to going "beyond phenomena." The event had been researched, prepared, staged, and was
intended to play the part of a "cause" in the history of physics
as such, and not only as part of the content of physical understanding. And yet, it was a genuine event, not. a "purely human"
operation, analogous to a convention or a decision that, because
produced by humans, could be called into question by them.
And the majority of physicists who had gambled against atoms,
against the legitimacy of hypotheses that invoked a discrete,
unobservable reality "beyond phenomena," acquiesced. Even
if they were the losers in this game, they couldn't use this as a
pretext to retaliate. It was as if "reality" had accepted the offer,
had agreed to the rendezvous proposed by those who intended
to impose the existence of atoms on everyone.
When, at the turn of the century, the young jean Perrin
decided to study colloidsemulsions that the newest generation of microscopes enabled scientists to describe in terms of
suspended grainshe was fully aware of what he was doing.
He was trying to create a new kind of experimental device, one
that allowed the physicist to demand that a discrete reality

14. A T O M S

ATOMS e x i s t !

EXIST!

*5

hypol hrtically situated "beyond" an observable phenomenon

eveiyone who, at the time, believed they could reduce physics to

assume its place as the only reality capable of explaining the

manipulative recipes, which, although convenient, were inca-

measurements directed at it. Until then, one spoke of atomic

pable of accessing reality.3

models, which meant that in spite of the fecundity of the atomic

In other words, the column of emulsion whose grains Per-

hypothesis, it was unable to silence those who considered it

rin is laboriously learning to count, and which are themselves

superfluous, dangerously irrational, or simply heuristic. Yet,

laboriously prepared so they are of equal size, not only modifies

"the only reality capable" reflects an ambition to provide atoms

the scale of atmospheric phenomena, it also modifies the scale

with a mode of existence that, following Bruno Latour, I have

of contemporaiy controversies. Among them, those that divide

associated with "experimental factishes"beings that we are

physicists over the vocation of physics and those that pit "ideo-

compelled to describe as having been constructed by us and

logues" against one another concerning the truth-value of the

simultaneously endowed with an autonomous existence.

For Perrin, a column of emulsion (where the density of


grains of gum is a function of height) is a "miniature ponderable
atmosphere; or, rather, it is an atmosphere of colossal molecules, which are actually visible. The rarefaction of this atmosphere varies with enormous rapidity, but it may nevertheless be
perceived. In a world with such an atmosphere, Alpine heights
might be represented by a few microns, in which case individual
2

atmospheric molecules would be as high as the hills." The challenge is obvious: if the column of emulsion indeed obeys the gas
law (where the density varies as a function of atmospheric pressure), it will serve as an explicitly kinetic model of gas behavior,
as defined by this law, a model that explains behavior in terms
of molecular movement. Additionally, the device can be used to
manipulate the "atmosphere" because Perrin can vary the viscosity of the liquid, the temperature, and the size of the grains
of gum. This should allow him to assign a value to "Avogadro's
number," the "N" that appears in every model in which "microscopic" actors and events are introduced. In other words, Perrin's manipulable atmosphere can create the difference between

sciences. Molecules become colossal in size and great ideas are


associated with a specific challenge, the possibility of quantifying Avogadro's number. The column of emulsion transforms
the relationships between ideas and things by inventing the
means to make the postulated invisible visible. Its purpose is to
obligate all physicists and, beyond them, anyone who might be
interested in the question, to recognize that atoms, ions, molecules, and radioactive nuclei are all endowed with an autonomous existence and have the means to resist the claim that they
are a product of human, and only human, interpretation. In
other words, Perrin s demonstration implies and conjoins what
are customarily distinguished as the language of physics and
the language of the epistemology of physics. Perrin succeeded
in coaxing atoms to bear witness to their numbers. In so doing,
they not only quantify Avogadro's number, they affirm the legitimacy of seeking, beyond the phenomena we can measure, a
reality that no one can reduce to a simple interpretation of those
measurements, a reality that cannot be convinced of being relative to our measurements, but one that must be recognized as
having the power to explain them.

the physics of phenomena, limited to measurements directed at

Yet, here, measurement has little to do with the measure-

observable macroscopic behavior, and the physics that inter-

ment that gave such prestige to mechanics, centered around the

rogates, through those phenomena, the invisible multitude

Galilean "equal" sign, and distributing the relative identities of

of microscopic actors. And in doing so, it will have silenced

cause and effect (see Cosmopolitics, Book II). The measurement

14.

ATOMS

EXIST!

ATOMS

EXIST!

*6

that allows atoms to bear witness is not relative to balancing

own rationale, or the definition of a state capable of communi-

something against something else but to detection. The micro-

cating instantaneous description and evolution over time. Such

scope enabled Perrin to observe the grains of emulsion and count

measurement could lead to a physics of active rationalization,

them at different heights. Other devices, such as the Wilson

similar to Pierre Duhem's, where the construction of a measur-

cloud chamber, using cascade effects, would soon amplify the

able object deliberately took precedence over realist ambition,

events that were being identified. Today, high-energy physics is

as well as to the dream of a physical reality "rational in itself,"

primarily a physics of detection devices, which sort and amplify

measurement of which would be limited to revealing its prin-

the highly rare events postulated by theory. In one sense, we

ciples. The object of Hamiltonian dynamics is, in one sense, the

might be tempted to say that the change of scale brought about

most integrally constructed of physical objects: all the variables

by the column of emulsion transformed Perrin into Maxwell's

that define it, far from corresponding to any direct observation,

demon, capable of observing a reality that is invisible to us (see

are themselves defined as a function of the system. And yet it

Cosmopolitics, Book III). But this is a risky analogy, for the demon

provides us with the most accomplished figure of an "autono-

did not merely observe, it brought to bear on microscopic reality

mous" object we have, one that is ceaselessly self-determining,

the requirement that cause be conserved in effect. Its manipu-

excluding any judgment that might be associated with a par-

lations corresponded to a knowledge of the gaseous milieu as a

ticular point of view. For it imposes a systematic relationship of

"dynamic system." For Perrin, the column of emulsion was not a

reciprocal translation on all representations that, by conserving

dynamic system. It constrained him to a practice that was some-

the Hamiltonian, confirm causal measurement and are there-

what analogous to that of the naturalist. He had to prepare his

fore defined as being "equivalent." To the autonomous object

approach, the device that allowed him to count the grains with

corresponds a physicist-judge, who knows how to interrogate

maximum precision. He then had to wait-wait for the grains to

the object, for she knows the principles to which that object cor-

become still, and for the emulsion to reach equilibrium. Today,

responds. She is an ideal figure of justice, devoid of all arbitrari-

at the bottom of mineshafts and beneath mountains, physicists

ness, for the categories of law that arm the judge are, at the same

have installed metal plates sandwiched with numerous detec-

time, the principles governing the actions of the accused.

tors. They too are waiting, this time for one of the protons in
the metal plates to demonstrate, through its disintegration, the
validity of the theories that assign it a finite lifetime. Although
superaccelerators accelerate the mad circular race of elementary
particles, they merely succeed in creating the conditions liable
to produce the desired particles through collisions. Once those
conditions are established, they too must wait, without knowing
when or where the events their detectors lie in wait for will be
produced.

As for measurement by detector, it is partial by definition.


It is not the actor as such who becomes observable. What the
device makes observable is theoretically associated with the
actor but generally does not enable us to identify that actor as
an object, that is to say, does not enable us to judge it. That is
why Perrin can forget the numerous differences between his
grains of emulsion suspended in a liquid and the nearly freely
moving molecules that, following kinetic theoiy, constitute a gas
such as our atmosphere. For Perrin, neither the grains nor the

To Galilean measurement's "full cause and total effect" cor-

molecules need to be defined as wholly intelligible objects. He

responded the construction of an object capable of defining its

withdraws only the information capable of confirming, and then

14.

ATOMS

EXIST!

cxploil mg, I lie validity of the atmospheric model represented by

ATOMS

EXIST!

*9

resemble a misuse of power. We would then have to acknowl-

I he colloidal solution. More generally, the detector is addressed

edge that, in general, the judge questions the accused in a lan-

to events insofar as they are producers of information about

guage the latter does not understand, and is in no way concerned

entities to which those events "happen," which implies that the

with finding out how the accused might describe what the judge

device is dependent on an event it can neither judge nor control,

would define according to the categories of the "crime." The

only await or provoke. Correlatively, the device is not "causal"

requirement is satisfied by the phenomenon, of course, but we

but "logical." It does not introduce a cause that is exhausted in

can also say that this satisfaction has been extorted. The judge is

some measurable effect it produces but a chain of implications

no longer authorized by reality.

and consequences enabling us to follow traces of the event back


to the actor that experienced or provoked it.

That is why Duhem argued that greater attention should be


given to the conditions of relevance of the different theoreti-

We must now distinguish two concepts of measurement and

cal constructs. Thus, the case of "false equilibrium" (think of

the reality to which measurement is addressed. This distinc-

an apparently stable gas mixture that a mere spark can suddenly

tion is the price paid by physics's new identity, its vocation to

ignite) challenges the relevance of the simple identification

encounter a reality "beyond phenomena."

between a chemically inert state and a state of equilibrium. Can

As we have seen, causal measurement is intended to con-

we use as a guide the analogy between "false equilibrium" and the

struct the object's identity, to authorize an objective judgment,

stone that, because of mechanical friction, remains motionless

but it also creates the question of the relevance of that judgment.

on a slope until a flick of the finger causes it to tumble downhill?

Clausius's "rational" thermodynamics showed that it was not

And how can we modify the edifice of thermodynamic equilib-

impossible to require that phenomena be presented in a way that

rium to integrate the chemical analogues of "friction"? How can

gave meaning to causal measurement, although satisfying that

we generalize the concept of chemical potential so as to enable

requirement did not legitimize its relevance for phenomena.

it to deline false equilibrium and the "chemical friction" that

Measurement, in this case, assigned full responsibility for the

corresponds to it? 4 These are the kinds of stark questions that

presentation to the one who imposed it. The phenomenon was

Duhem claims are relevant, questions that obligate the physi-

subject to a requirement through which it could be judged, but it

cist-judge to proceed with tact and skill, to negotiate the "cases"

did not provide confirmation that the judge was impartial. This

before her according to her own rules, but not blindly, without

leads to the following question: Wouldn't mechanics be merely a

confusing those rules with a language that reality might speak.

particular case, and thermodynamics a lucid generalization? In

The question of the relevance of the judgment that con-

effect, mechanics would be the only case where the relevance of

structs an object can even go so far as to challenge the funda-

judgment could be confirmed by the object under investigation,

mental notion of trajectoiy found in dynamics. In The Aim and

the only case where the "remainder," the deviation between the

Structure of Physical Theory, Duhem affirmed the limits of this con-

mechanical system and the ideal being that would completely

cept even when the equations that appear to guarantee its legiti-

satisfy the requirements of measurement, could be reduced to

macy can be constructed. At the time he was nearly the only one

"secondary" factors such as friction. In the general case, judg-

to draw radical conclusions based on the identification of the

ment, if it claimed to have the authorization of its object, would

first "pathological trajectories," trajectories that specialists in

io

ATOMS

ATOMS

EXIST!

dynamics today refer to as fractal or chaotic, and that are now


associated with the current "renewal of dynamics."

EXIST!

11

in 1906, he quite clearly described the dynamic behavior now


known as chaotic, characterized by a blend of qualitatively dis-

For Duhem, trajectory had to be abandoned whenever it

tinct trajectories regardless of the scale. Moreover, Duhem

failed to satisfy the requirements its use assumedin this case,

foresaw the fecundity of the problem. Although Hadamard s

whenever its mathematical representations ceased to be robust.

mathematical description signaled for him an irreconcilable

Trajectory is an instrument of mathematical deduction and

divorce between the mathematical entity and physical repre-

"a mathematical deduction is of no use to the physicist so long

sentation, which had to be robust, which demanded a "math-

as it is limited to asserting that a given rigorously true proposi-

ematical representation of approximation," this divorce was not

tion has for its consequence the rigorous accuracy of some such

synonymous with the abandonment of mathematical construc-

other proposition. To be useful to the physicist, it must still be

tion. It created the challenge of having to construct a new math-

proved that the second proposition remains approximately exact

ematical representation that was physically pertinent: "But let us

when the first is only approximately true. And even that does

not be deceived about it; this 'mathematics of approximation is

not suffice . . . it is necessary to define the probable error that

not a simpler and cruder form of mathematics. On the contrary,

can be granted the data when we wish to know the result within

it is a more thorough and more refined form of mathematics,

a definite degree of approximation." 6 What the mathematician

requiring the solution of problems at times enormously diffi-

Jacques Hadamard defined in 1898 is a situation that makes

cult, sometimes even transcending the methods at the disposal

deduction unusable. Take a complicated surface, such as the

of algebra today."8

forehead of a bull whose horns and ears proceed outward toward


infinity. We can draw a particular geodesic curve on that surface:
"after turning a certain number of times around the right horn,
this geodesic will go to infinity on the right horn, or on the left
horn, or on the right or left ear . . . No matter how we increase
the precision used to determine the practical data, reduce the
size of the spot where the initial position of the material point
is found, or tighten the bundle that contains the velocity's initial direction, the geodesic that remains at a finite distance by
turning continuously around the right horn can never be freed
of its unfaithful companions, which, after having turned around
the same horn with it, will deviate infinitely. The only effect
afforded by this greater precision in establishing the initial data
will be to require that the geodesies describe a greater number
of turns around the right horn before producing their infinite
branch. But this infinite branch can never be eliminated." 7
I have cited Duhem at length to show that, in a book published

Therefore, Duhem's perspective has nothing to do with that


of a tranquil "instrumentalist," who claims to limit herself to
the information provided by her measuring instruments. While
"nothing is true in itself other than the results of experiment," 9
mathematical physics cannot be purely and simply skeptical, for
it has to be relevant. It is a Pascalian situation explicitly transposed to physics: "We have an impotence to prove invincible
by any dogmatism, and we have an idea of truth invincible by
any skepticism." 10 The solution to the dilemma is found in an
ethics of obligation. To the extent that the physicist resists the
"needs of the imagination" and fully accepts the twofold obligation imposed by mathematical consistency and experimental
relevance, she may hope to create a theory capable of systematically translating phenomenal variety. It is then, and then only,
that the order in which theory categorizes its results might lead
us to conclude "that it is or tends to be a natural classification. "ll
Like any real physicist, Duhem can thus claim that "the belief in

14.

ATOMS

EXIST!

ATOMS

EXIST!

*3

an order transcending physics is the sole justification of physi-

that a particular species of "Man," the physicist who goes beyond

cal theory." 12 The difference is that, for Duhem, physics can be

phenomena, can now hope to approach through detection and

true to its purpose only to the extent that it avoids any confusion

numbers.

between what it constructs and what transcends it.

When Perrin's figures made their appearance, there disap-

It is the ability to identify and judge, inseparable from the

peared from view the "transcendent" order Duhem hoped the

language of rational mechanics and thermodynamics, that has

edifi.ce of physical theory might allude to. "Reality" seemed to

conferred its value as a critical concern to the question of rel-

invade the physicist's laboratory, dictating figures and more fig-

evance and its corresponding obligations. It is the judge, rather

ures, which fed the most extravagant appetite for discovery. But

than the hunter or naturalist lying in wait, who must ceaselessly

it is at this point that we must pay close attention. The appetite

bear in mind that the crime defined in terms of legal categories

that was fed was no longer that of a judge. The "reality" to which

must not be confused with the act that has led the accused before

measurement by detection is addressed is assumed to be auton-

the court. That is why jean Perrin had no problem with "reality."

omous, but is no longer defined by the convergence between

And at the time he succeeded in calculating Avogadro's number,

description and reason. The physicist's imagination, which real-

he even took great pleasure in literally wallowing in the num-

ity authorizes, contrary to Duhem, is addressed to beings that

bers that designate the microscopic reality whose autonomous

allow themselves to be described but do not give the describer

existence he was now able to confirm: "In short, each molecule

the power to identify them, or subject them to a law. Whereas

of the air we breathe is moving with the velocity of a rifle bullet;

causal measurement creates both an object and the judge of that

travels in a straight line between two impacts for a distance of

object, providing the "equal" sign with the power to subject a

nearly one ten-thousandth of a millimeter; is deflected from its

phenomenon to a judgment, measurement by detection recog-

course 5,000,000,000 times per second, and would be able, if

nizes in what it addresses an "activity" that does not dictate how

stopped to raise a particle of dust just visible under the micro-

it should be described. Detection does not demand the power of

scope by its own height. Three thousand million of them placed

judgment, only the ability to characterize unequivocally.

side by side in a straight line would be required to make up one


millimeter. Twenty thousand million must be gathered together
to make up one thousand-millionth of a milligram."' 3 These figures all result from substituting Avogadro's number, N, by its
value in the kinetic model of gases, and they satisfy the imagination of the physicist beyond all expectations. A Pascalian emotion, like Duhem's, gripped Perrin: "The conclusion we have
just reached by considering a continuously diminishing center can also be arrived at by imagining a continually enlarging
sphere, that successively embraces planets, solar systems, stars,
and nebulae. Thus we find ourselves face to face with the now
familiar conception developed by Pascal when he showed that
man lies 'suspended between two infinities.'" 14 Two infinities

However, we cannot say that the requirement concerning


reality has diminishedit has been transformed. To be valid,
detection requires that the information produced has an unambiguous meaning, and that it can be associated with an actor. But
the main point is that this requirement has become "nomadic."
Unlike causal measurement, it does not singularize physics. Pasteur succeeded in transforming epidemic disease and fermentation into detectors of the presence of microorganisms. In order
for disease to be recognized as able to "detect," he had to be able
to demonstrate that different diseases are caused by different
microorganisms and that each kind of microorganism results
in a single disease. However, he did not have to prove that the
microorganism constituted the "rationale" of the disease, the

14.

ATOMS

ATOMS

EXIST!

EXIST!

*5

way in which the disease had to be understood.' It only had to be

without judging the culpable. Here, "object" celebrates the

a "necessary condition." In other words, the "experimental fac-

construction of a coherent representation of the actor respon-

tish" constructed by detection answers the questions addressed

sible for everything detected about it. Between the object and

to it, but its responses, although veridical in the sense that they

the "reality" it designates, there exists an irreducible distance,

resist the accusation of being mere artifacts, are still not "truth-

separating detection and identification, one that new methods

ful" if by this we mean the avowal a being can be led to make

of detection may be able to reveal but not reduce, clearing the

concerning its own truth.

way for the discovery of new responsibilities.

Of course, measurement by detection hardly implies that

The object of dynamics guaranteed the exhaustive nature of

the physicist is a pure observer, which is sometimes the posi-

dynamic description because the cause, through its equivalence

tion of the naturalist. Like causal measurement, it can provide

with effect, guaranteed the vanity of any attempt to describe

the scientist with the power to characterize its objects in terms

"differently." The same is not true of the representation result-

of well-defined variables and construct functions. Activity can

ing from detection. For this relates the causal order to its origi-

be modulated, the detector can filter according to various crite-

nal register, where cause is understood in several ways and

ria. But, in all cases, detection assumes that "reality" takes the

supports nothing other than logical reasoning. The microor-

initiative, that it is endowed with the ability to act. It has become

ganism has the power to cause disease, true, but this power may

"autonomous" in a way that differs completely from the dynamic

vary in the laboratory, where living beings have become either

object. The dynamic object is simultaneously and identically

reproductive media or instruments for detecting the virulence

autonomous and submissive, "obeying its own laws," whereas the

of the being under investigation.' 6 But in spite of the accumu-

detected actor can, depending on circumstances, be considered

lation of propositions concerning this power, they themselves

"spontaneous," comparable to "natural" radioactive disintegra-

are not sufficient to identify the cause, the microorganism, and

tion, whose rate of decay nothing can accelerate or decrease, or

the effect, the disease. On the contrary, they tend to multiply the

subject to laws to which detection does not provide access.

problems presented by these two terms. This is the key differ-

In one sense, we could say that detection implies a more

ence between cause in the sense of dynamics and cause in the

"realist" approach than causal measurement, because it subjects

logical, or syllogistic, sense. Only the first has in itself the power

itself to a reality it does not claim to reduce to a rational iden-

to define its effect, for its identity is nothing other than this

tity. Perrin's atoms, like Pasteur's microbes or the atomic nuclei

power; the second is relative to detection and its interpretation,

that disintegrated over Hiroshima, exist "truly" in the sense that

and other questions, other detection techniques can transform

things "happen" to them and that they cause things "to happen,"

it without ever achieving anything other than logical closure.

in the sense that they affect their environment, to which the

Closure that is still liable to neglect unidentified ingredients or

detector belongs, and can be affected by it. But here "reality" has

conditions, or to dissimulate the fact that the "cause" presented

lost the power to dictate how it must be described. We can, of

to us is liable to belong to quite different narratives under other

course, by extension, still speak of an object, but we must then

conditions. Pasteur's microorganisms are not arbitrary, but their

overlook the legal connotations associated with the term and

identity as a "cause of disease" is eminently variable. Whereas a

borrow the vocabulary of inquiry, distributing responsibilities

dynamic "cause" imposes its "ontology," that is, its definition of

14.

ATOMS

EXIST!

ATOMS

EXIST!

*6

what a cause is, a logical "cause," according to Latour's expres-

that allow it to exist? Haven't physicists finally abandoned the

sion, is "ontologically variable": not only what it causes but the

dream of knowledge that is able to access a reality that disquali-

way it causes are subjects for histoiy.

fies all other forms of knowledge? In fact, as we shall see, this

The reality quantified by Jean Perrin vanquished Duhem's

interpretation could have been legitimate. It would have been if

"transcendent" order. But it too could have communicated with

what I will here call the "lesson of Niels Bohr" had been heard

a form of transcendence, that of a world where responsibility

and accepted. And in that case, this entire exploration of phys-

cannot be used to identify the responsible party, that is, to pen-

ics would have been pointless. Physicists would today introduce

etrate its reasons. Which it did not do. And it is possibly because

themselves nonpolemically, as vectors of lucidity, if not as vec-

it did not do this that the "physical realism" Duhem had tried to

tors of peace among fields of knowledge. Which is not exactly

disrupt was finally discovered to be in a state of crisis. The prob-

the case. The very expression "the real is veiled" is hardly one of

lem posed by Duhem may well have been part of an outmoded

peace. For here, the veil is not something a practice that brings

past, outmoded by the intentional and successful ambition to get

actors into existence would create in the very act in which it

"beyond phenomena"; nonetheless, it effected a rather strange


return in a different form, the much more paradoxical and fascinating form given to it by quantum mechanics.

"unveils." It is not there to signal, for example, the difference


between the identification of responsibility and the identification of the one responsible. Veiy specifically, the veil is what

One of the aspects of anamnesis I attempted to develop

specialists of quantum mechanics alone have to confront; it is

throughout Book III was the break with the idea that the revo-

the new, dramatic message of physics. What is celebrated is the

lutions of twentieth-centuiy physics had disturbed the tranquil

disappointment of physics's leading ambitionsthe impossibil-

course of "classical" physics, that is, a physics ruled by a form of

ity of satisfying the physicist's vocation, access to a unique and

ahistorical progress. Correlatively, here anamnesis is no longer

intelligible vision of reality. Therefore, the veil consecrates the

necessary. The "crisis" of quantum physics and the "problem of

vocation whose impossibility it proclaims.

the interpretation of quantum formalism" belong to contempo-

And thus, the "reality-for-us," which is relative only to

rary culture. And I will approach them as such, as subjects of dis-

"communicable human experience," designates neither an "us"

cussion rather than as more or less uplifting forms of pedagogy.


I will begin by introducing a rather startling contrast: in 1913,
physicists, whether enthusiastic or resigned, acknowledged that
the world was indeed a confusion of atoms. Yet, less than fifteen
years later, when those atoms were part of theory, theory led its
interpreters to assert that the "real is veiled" and we can only
access an "empirical" reality, which is to say, a reality-for-us.17

nor an "experience," nor a "communication," which by that


very fact might become interesting in their multiplicity, but an
act of mourning that confirms the vocation of the physicist by
announcing the impossibility of its fulfillment. And this proud
vocation is still very much alive whenever a physicist feels justified in making human consciousness an inseparable correlate of
what "we" nonphysicists "naively" refer to as an "independent"

Some may feel that the fact that specialists of quantum

reality. The exploration of the questions of quantum mechan-

mechanics speak of a "reality-for-us" should be a cause for

ics, therefore, becomes part of my "cosmopolitical" project on

rejoicing. For, doesn't it show that the physics they represent

the following basis: how can we transform the "mourning" that

has finally been recognized as being relative, like all other areas

some physicists feel they can assume on behalf of all humans

of knowledge, to the questions, requirements, and obligations

into a "history of physicists"?

ABANDON

THE

DREAM?

J9

nature of the respective requirements and obligations of its


various actors, makes it opaque from a point of view other than

thatalways situatedof its actors. In other words, there is nothing contingent about the lack of discussion concerning what is

Abandon the Dream?

at play; it has nothing to do with the laziness of commentators or


the competitive urgency of physicists. It is as if the only aspect of
the physicist's vocationfinding a single intelligible representation of the physical worldthat has remained truly invariant
and explicit is the ambition to get "beyond," and in this case not
only "beyond" the diversity of phenomena but beyond the ordinary conditions of experimentation. But this "beyond" comes
at a price. For, whenever physicists today maintain the values
of physics, it is above all to encourage us to accept the massive

Some people find it surprising that the problem of quantum


mechanics remains a vital one even today. After all, quantum

financial investments their equipment requires.


It is not impossible to tie this practical opacity to the current

formalism is roughly sixty or seventy years old and, since then,

strange proliferation of statements from physicists who assume

quantum field theory, high-energy physics, and the unification

the role of thinkers and chroniclers of future revolutions. As

of fundamental interactions have transformed physical repre-

if anticipating increasingly scandalous future revolutions held

sentation. Is the survival of a long-standing debate symptomatic

the place formerly occupied by the production of intelligibility.

of the laziness of commentators, who limit themselves to things

As if the rush toward such dizzying perspectives might enable

they can "understand"? Or is it the result of the sense of com-

them to transform the problems created by physical theory into

petitive urgency felt by the new generation of physicists, who no

the promise of new triumphs. In today's speculative physics, the

longer bother to take the time to elaborate or explain the nature

effects of a possible revolution, far from being unexpected, are

of contemporary physics to outsiders? My position is that there

often already taken into account and exploited. So, for example,

is indeed a symptom, but this symptom may reflect an entirely

in The Emperor's New Mind, Roger Penrose, discussing the prob-

different problem. Physicists representative of the older (and

lems of interpreting quantum mechanics, introduces the "germ

already obsolete) new generation, such as Richard Feynman,

of an idea," which he uses to announce a "revolution" that is

directed streams of irony at those who believed there was still

without doubt going to be more important than that of relativ-

something to "understand" in physics. Naturally, there are a

ity or quantum mechanics. 1 ' Revolution" is no longer an event,

great number of things to learn about the practices of contem-

it is a component of the physicist's management of future per-

porary physicists, especially concerning the difficult, strained

spectives. In Penrose's case, the problems raised by quantum

relationships between experimental physicists, phenomenolo-

mechanics will be resolved when gravitation has assumed its

gists, and theoreticians that characterize high-energy physics.

legitimate place.

But the very complication of the freld of practice, the obscure

18

This way of anticipating the future to settle the problems

15 A B A N D O N

THE

DREAM?

ABANDON

THE

DREAM?

27

of the past is, to say the least, surprising. As a theoretical -

why that moment resulted, in one way or another, in the phys-

experimental science, quantum mechanics is recognized for

ics of the "new generation" silencing those questions, indulg-

its impressive success, that is, for the precision with which the

ing in ironic skepticism and unbridled speculation. "No one,"

data actually produced by experimental means have confirmed

wrote historian Stephen Brush, "has yet formulated a consistent

its predictions. The fact that these devices have not assigned

worldview that incorporates the Copenhagen Interpretation of

gravitation a crucial role implies that no one has yet invented

Quantum Mechanics while excluding what most scientists would

a way for gravitation to manifest the consequences of its exis-

call pseudo-sciencesastrology, parapsychology, creationism,

tence in this experimental context. The idea that it would play

and thousands of other cults and doctrines." 3

a rolea role both secret and crucial because failing to account

What is this "Copenhagen interpretation" that Brush alludes

for it would be mysteriously responsible for the singular formal

to? For some, this interpretation is that of Niels Bohr, who spent

structure of the theory and the problems of interpretation this

his whole life thinking about the implications of the physics

structure entailsamounts to claiming for this physical-math-

he had helped redefine. For others, it is the "orthodox" inter-

ematical language a kind of authentically transcendent truth.

pretation of quantum mechanics, one that managed to displace

And that truth would be such a truthful rendition of reality that it

all the others. There is an alternative view, however, although

would have the ability to manifest, through a quirk of its formal

not shared by many. It entails, on the one hand, a reference to

structure, that a dimension of that reality is lacking.

a relationship that history failed to smooth over, to trivialize,

It is not impossible to maintain that an already dated quan-

between a man and a scientific innovation, and, on the other,

tum mechanics preserves its attraction for the simple reason

a consensual orthodoxy that imposed itself without having the

that the physics that followed is marked both by the active rejec-

ability to convince anyone, as if history, in this case, no longer

tion of any idea that is not a simple "reflection of the state of

had the power to silence "vanquished" objections and frame the

the question as defined by physicists" and by wild speculation

controversy in the victors' terms. As if reference to the "van-

uncoupled from practice, adopting as its only guideline a ref-

quished" in the controversy over quantum mechanics lingered

erence to the revolution that makes conceivable for tomorrow

on, along with the idea of a possible alternative.

what is inconceivable today. Nor is it impossible that this dual

In fact, the history of quantum mechanics does not respond

singularity, the practical opacity of so-called high-energy phys-

at all to the canon of the history of science. The anomaly is all the

ics and the grandiose perspectives it has been associated with,

more striking given that, from the point of view of the produc-

might be analyzed in terms of an "ecological isolate." It may pos-

tiveness of the relationship between theory and experimenta-

sibly be related to the disappearance of any necessity, indeed,

tion, of its ability to encompass all fields of physical chemistry,

any possibility, of establishing relationships, even conflicting

and the creation of new methods of observation and measure-

ones, with other practices that address the "same" reality.2

ment, it fulfilled its creators' most ambitious hopes. But it is as if

In any event, the quantum mechanics of old remains a cru-

this successful history was also a source of permanent intellec-

cial reference. It marks the moment when physicists once again

tual dissatisfaction, one a multitude of authors have attempted

debated what they were able to require of reality and the obliga-

to counteract.4 The term "orthodox," a very unfamiliar one for

tions imposed by that reality. And it is important to understand

scientists, concentrates the ambivalence of the situation. For

22

ABANDON

THE

DREAM?

ABANDON

THE

D R E A M ? 27

those who feel that the success of quantum mechanics consti-

denounced as the "curse of positivism" by LeonRosenfeld, Bohr's

tutes its necessary and sufficient source of legitimacy, the term

disciple and spiritual heir. 5 In fact, the rejection of positivism

refers to an interpretation that is external but useful because it

was the one thing Bohr and Werner Heisenberg most strongly

silences the futile problems that others insist on presenting. For

agreed on. "We must then ask why Bohr and Heisenberg were

those who refuse to acknowledge the situation and attribute its

unable to escape the "curse" in question. Why were they unable

longevity to the triumph of a mediocre, instrumental concept,

to get others to accept what they defined as the obligations of

through which physics has abandoned thought, it sounds, on the

quantum mechanics, becoming instead, and quite unwillingly,

contrary, like a protest and an accusation.

for everyone who saw prohibitions rather than obligations, the

This situation is even stranger when we consider the differ-

authors of the "orthodox" interpretation of quantum mechan-

ence between Niels Bohr's own interpretation and the "orthodox"

ics? Here, as elsewhere, general ideas, determinism or realism,

interpretation attributed to him, which frequently assimilates

risk masking the practical question tying together the problem

the very distinct ways of thinking of Bohr and Heisenberg. In

of values and the technical definition of problems, the question

fact, these stylistic nuances seem superfluous to anyone who

that helped to consolidate the dissatisfaction of the opponents

refers positively to the so-called orthodox interpretation. For

of the Copenhagen interpretation.

them, this interpretation is reduced to the authorization to

To address this practical question, I'll begin with a contrast.

move forward without asking pointless questions. Quantum

Critics and partisans alike of the Copenhagen school celebrate

formalism would present no problem; it would simply obligate

Einstein's theory of relativity as a model. Bohr and Heisenberg

physicists to abandon the dream that caused them to confuse the

have always said they have faithfully followed the example of

information supplied by their instruments with the attributes of

their older peer. Why have critics not acknowledged their fidel-

a given reality. Quantum mechanics would be nothing but a "for-

ity? Why did relativity escape the "curse of positivism"? Why, for

mal machine" that produces predictions about observable data

some, do relativity and quantum mechanics represent a contin-

on the basis of observed data. The purely instrumental nature

uously coherent history and. for others (including Einstein), a

of this knowledge would be a point of no return because what it

great histoiy interrupted by betrayal and abandonment?

asks physicists to renounce would never be anything other than


an illicit dream anyway. Correlatively, the critics of the orthodox
interpretation have no need for stylistic nuance. It is against the
threat of a positivist and instrumentalist reduction of physics
that they defend the vocation of physicists, the faith in the intelligibility of reality that animates them.

Bohr and Heisenberg believed they were following Einstein's


example. The argument that allowed Einstein to deny that the
concept of the absolute simultaneity of two distant events had
a physical meaning was based on the fact that no experimental
device would ever allow observers moving relative to one another
to agree on this simultaneity. And Bohr based his claim that

However, it is not unanimously held that the "orthodox"

quantum formalism constituted a "complete" theory on the same

interpretation can be expressed in "positivist" terms, that is,

type of argument: to assign a physical meaning to an element of

a general epistemological claim that rejects, in principle, any

theoretical description implies that conditions of observation

attempt at intelligibility that is not limited to the coherent inter-

can be defined in relation to which that element can acquire its

relation of observable data. This kind of presentation was even

meaning. Moreover, it was at this point that he "awaited" those

25 A B A N D O N

THE

DREAM?

ABANDON

THE

DREAM?

27

who hoped to get "beyond" quantum formalism. He was easily

time.7 But this responsibility provides the observers with no

able to demonstrate to them which of their hypotheses brought

more than a subordinate role, defined by the fact that measure-

into existence a "beyond" that, like simultaneity at a distance,

ments in space and time separate what, in truth, is inseparable.

can never be subject to any possibility of measurement. Why,

Thus, the world of our obseivations is certainly only phenom-

then, did Einstein allow himself to be caught in Bohr's trap, and,

enal, but we are not prisoners of this phenomenal world. We are

more crucially, why did he resist the outcome, when he was the

able to judge it in the name of four-dimensional mathematical

one who had invented both the trap and the outcome?

I ruth: according to the terms of this judgment, we know how our

We know that if, for a time, the Viennese positivists who

space and time observations indirectly testify to the existence of

followed Ernst Mach saw in Einstein's theory of relativity the

space-time. The presentation of this judgment becomes a sign

triumph of their claim, Einstein successfully severed any con-

of gloiy for a physics that managed to understand the relativity

nection to these early allies and relativity escaped the stigma of

of appearances and uncover the absolute from which they are

the "curse of positivism." 6 Relativity was successful in showing

derived. Laplace's observing demon is dead, but the Queen of

that it did not satisfy a general obligation of an epistemological

Heaven continues to reign as magisterially as ever, her incom-

nature, which ordered scientists to eliminate any type of resem-

municable attributes now being identified with a mode of math-

blance between their theories and the speculative descriptions

ematical existence that can only be subject to communication,

adopted by metaphysics. Relativity's obligations were unique,

that is to say, measurement, at the price of a redefinition of phe-

associated with the unique properties of the speed of light, prop-

nomena. These phenomena resemble the phenomena in Plato's

erties that condemned, in spite of the intuitive evidence, the

cave, but now it is the physicist who conceives the mathematical

possibility of providing a meaning to simultaneity that was not

truth of the "Sun," placing the distorted shadows in the correct

relative to the observer. Moreover, those obligations did not

perspective.

result in the abandonment of an "objective" description dictated


by the object but in the reinvention of that objectivity. For physicists, relativity is not positivist because it does not question
the possibility of a description that is independent of observation. It "merely" obligates us to recognize that that description
cannot be spatiotemporal. Objective description has the status
of a matrix for all possible spatiotemporal descriptions, which
depend on the frame of reference of the one who describes
them.

Quantum mechanics, on the other hand, centered on


Planck's universal constant, appears to preclude the dream of
an "objective" description of quantum beings. But it does worse
than this. For the formalism that precludes the dream also succeeds in giving rise to it. As I'll show later, it provokes the dream
so that it can preclude it later on. In other words, it both "disappoints" and "deceives."
It is "disappointment" that serves as the customary thread
in stories about the creation of quantum mechanics. The start-

The universe of general relativity is populated with observers

ing point emphasizes the continuity between classical dynam-

who conduct measurements of simple registration, armed with

ics and quantum mechanics, and the central role played by the

chronometers and standard meters. It is they who are respon-

"wave mechanics" associated with Louis de Broglie and Erwin

sible for the transition from an "objective" four-dimensional

Schrodinger. Wave mechanics purported to be the worthy heir

mathematical description to a physical description in space and

of the Queen of Heaven and appeared to promise physicists an

ABANDON

26

THE

ABANDON

DREAM?

THE

DREAM?

27

authentic faiiy tale, a triumph of mathematical harmony that

interpretation given by Max Born, the reduction of the wave

was able to reconcile the science of masses in motion and the

function should be associated with knowledge production in the

science of light. It was at this point that an abrupt shift occurred:

strong sense, in the sense that what is "known" is transformed

de Broglie and SchrOdinger were "stripped" of their creation,

by that very production.

the "wave function" whose mode of behavior served as a splen-

In the conventional narrative, this is often the positivist

did match to the intelligible reign of the Hamiltonian Queen of

curse that Carabosse utters over the cradle of the wave func-

Heaven. It is almost as if the unwanted appearance of the wicked

tion that believes itself to be a Hamiltonian princess. This wave

faiiy Carabosse had transformed the young princess, already

function, Carabosse says, will disappoint all your hopes. You will

blessed by Einstein, into an ugly old crone. For we later learn

be forced to accept that it does not represent reality but merely

that the wave function doesn't describe any "physical reality." In

the predictive ability that describes the state of our knowledge.

fact, the wave function apparently doesn't describe anything at

Don't be surprised if it is transformed every time it is measured,

all; it simply allows physicists to predict probabilities, the prob-

this is simply the result of the fact that your knowledge and your

ability that the measurement of a physical quantity will have a

ability to make predictions change. Physicists were asked to

given result. The function itself corresponds to "probability

reject any reference to a reality that "existed independently of

amplitudes." 8

their production of knowledge."

Einstein's observers brought about a spatial and temporal

Such a statement is unquestionably a curse for the physi-

"localization" whose (mathematical) consequences "caused"

cist. Even Pierre Duhem would have refused to abide by it; for

distances, durations, forces, and accelerations "to appear." But

him, "reality" existed as a source of obligations, obligating one

this localization, the result of location measurements carried

to employ tact and skill in the use of theoretical instruments.

out by the observers, is incapable of affecting the "four-dimen-

The difference between Duhem's conception and that which

sional reality" that situates them and allows them to determine

quantum mechanics appears to celebrate in its positivist ver-

what will be observed. On the other hand, the transition from the

sion is the difference between the practice of a skilled artisan

wave function to observable probabilities establishes a disturb-

and the fantasy of a push-button device. The artisan does not

ing compromise that makes it impossible to situate "observable

know the material "in itself," independently of gesture and tool,

reality," to describe it as determined in principle by "physical-

but her knowledge is much more intimate than anyone else's,

mathematical reality." The "reduction of the wave function" to

through the relevant precision of her gestures and her choice of

observable probabilities cannot be assimilated to the definition

the appropriate instrument. .And it is such distinctions, Duhem

of a pathway to some "higher reality," indifferent to measure-

claimed, cultivated through intimate knowledge, that can claim

ment. Something "happens" to what is being measured. The

to express a "natural classification" that refers to an order that

wave function is not "reduced from the observer's point of view"

transcends physics. The push-button device, on the other hand,

but actually reduced; the representation of the quantum system,

relates to the fantasy of an all-powerful instrument of interroga-

which is to say, the probabilities resulting from previous mea-

tion, one that no one has refined or adjusted, and which creates

surements or the preparation of the system, are transformed

knowledge in spite of its complete contempt for the problem

after the measurement. According to the

posed by what it is we are tiying to know.

"disappointing"

28

ABANDON

THE

DREAM?

How is it that positivism, which can customarily intervene


only "afterward," in the form of a sempiternal reminder to
indifferent physicists, intimating that they should not confuse
"observable data" and "access to reality in itself," has been so
closely associated with a physical theory? Where did the positivist curse obtain its power?
In fact, the customary narrative strategy is rather curious, for the wave function seems to have come to us out of the
blue. The singular practices from which it may have originated
are relegated into the background. More specifically, it continues to be presented in terms of its relationship to the ambition
that de Broglie and Schrodinger believed it embodied. However, responding to the challenge of a theorization of the atom's
behavior was not an ambition de Broglie and Schrodinger associated with the wave function. Such a theorization existed already;
it was the "first quantum mechanics" initiated by Niels Bohr. De
Broglie and Schrodinger's ambition was to bring that first theory
into line, to avoid aspects of it they felt were scandalous.
In other words, the customary narrative strategy, which
related a hoped-for continuity between Hamiltonian physics and quantum mechanics, and the disappointment of that
hope, silenced or attenuated the fact that a rupture had alreadytaken place, brought about by the model of the atom proposed
by Niels Bohr. The hope that was disappointed by de Broglie and
Schrodinger's wave function is not the hope of establishing continuity between "conventional representation" and a representation of the atom, it is the hope of reestablishing a continuity
that had already been broken.
To escape the positivist curse, fairy tales, and Carabosse, it
might be worthwhile to tell the story somewhat differently, to
focus on the novelty of Bohr's model, of the way he redefined
the practical relationship of knowledge involving the physicist
and the new type of being known as the atom. For the scandal
that de Broglie and Schrodinger experienced, the break they had

ABANDON

THE

DREAM?

27

hoped to smooth over, can take on another meaning, one that is


fully realistic in Duhem's sense. For, it is because it was able to
constrain Bohr to bring about this rupture, to dare to create a
scandal, that the atom is indeed "real." The transformations of
the theoretical edifice physicists felt obligated to undertake in
order to produce its relevant description constitute its claim to
an existence that "transcends" physical knowledge.
Bohr's quantum model of the atom introduced electrons
arranged in discrete "stationary states" around the nucleus and
making "quantum jumps" by which, with no time having elapsed
and no space having been crossed in the sense understood by
dynamics, those electrons "jumped" from one state to another.
This discrete change of state could not be described, but it could
be defined by its energy cost, which measured the energy distance between orbital "states."
Bohr's atomic model, therefore, reverses the terms with
which the question of dynamic motion is presented. Dynamic
observables are spatiotemporal; energy is a constructed function, affirming the conservative character of any dynamic temporal evolution, its definition in terms of complete equality and
total effect. In Bohr's model, it is the energy cost of the nonspatiotemporal change of state that is directly observable. It corresponds to the quantum of light emitted or absorbed by the atom
during the change of state. Moreover, the states themselves
are defined only by means of those transitions, as terms of the
energy difference measured by light quanta.
In Bohr's model, we cannot observe an atom the way we
think we can observe the Moon, by visual identification. We
know atoms, in the sense that they are defined by the model of
stationary electronic states, only to the extent that they absorb or
emit light quanta, that is, the extent to which they interact with
their environment. Regarding this situation, Bohr was literally
indefatigable, speculating on the novelty of quantum mechanics
and the new obligations that followed for the physicist, even on

20 A B A N D O N

THE

DREAM?

ABANDON

THE

DREAM?

27

his deathbed. 9 But it is this "nonmechanical" model that Louis

l>eing" endowed with the power to impose new questions, in the

de Broglie and Erwin Schrodinger felt to be unacceptable. That

positive sense of the term. More specifically, it introduces this

is why they directed all their hopes to the construction of a wave

power in negative fashion: although endorsing it, it transforms

equation. The wave equations, when the appropriate limit con-

the problem. The quantum formalism created in 1937 promotes

ditions are used, have a discrete set of solutions. Therefore, they

the forgetting of obligations associated with the encounter with

might help eliminate the scandalous notion of an energy that is

microscopic reality, to the benefit of an apparently metaphysi-

not a continuous function of position and velocity, or help inter-

cal problem that has continued to fascinate philosophers ever

pret stationary states no longer in terms of discrete energy levels

sincethe problem quantum mechanics presents us with of

but on the basis of frequencies compatible with the limit condi-

having to choose what will become the object of measurement.

tions. The scandal of quantum "jumps" could then disappear to


the benefit of simple "frequency changes."
Wave mechanics was unable to tame the scandalous singularity of the behavior of the quantum being introduced by Bohr.
If we refer to the well-known distinction between "context of
discovery" (what the discoverer thought, believed, and hoped)
and "context of justification" (what subsequently happened and
can be substituted for the tale of discovery), the hopes of de Broglie and Schrodinger should have been forgotten, and the wave
function should have been presented as a new way, more powerful and more rigorous from the physical-mathematical point of
view, of theoretically acknowledging the new practice to which
the atom constrained the physicist. Yet, the customary narration
preserved the memory of the hopes and disappointments of the
creators of the wave function and is therefore oriented not to the
scandal of a reality endowed with the power to refuse to submit
to the requirements of a classical spatiotemporal description,
but to the disappointment of those who had attempted to restore
that submission.
However, that particularity of narrative, which helped make
quantum mechanics vulnerable to the "curse of positivism,"
leads to another particularity, one associated with theoretical
language itself. A very unique relationship binds Bohr's first
quantum model with the formalism we have inherited. The latter, contrary to the atomic model, does not introduce a "quantum

NIELS

BOHR'S

LESSON

33

wi ll k nown uncertainty relations, which present us with "com(ilnnentaiy" variables that singularize quantum formalism,

s nrliibles that cannot be measured simultaneously, and between


> I licit a choice must be made.

Niels Bohr's Lesson

This was not Niels Bohr's claim but Werner Heisenberg's.


Ami it is Heisenberg who would marshal the arguments around
which controversies concerning the relationship betweenphysIciil knowledge and reality currently revolve.
The idea of a perturbation implies reference to what is perturbed ("reality as it exists independently of measurement")
,iiid, therefore, immediately raises an objection by physicists.
Why, they asked, can't the reference to what is characterized by a
11 uantity of motion and a position be preserved, even if the per-

How should we think of measurement? According to the positiv-

turbation excludes the possibility that these can be measured

ist approach, measurement has no exterior, cannot be referred

simultaneously? However, those who objected were imme-

to a reality capable of defining "what" is being measured. From

diately confronted with what might be called the Einsteinian

this point of view, the impossibility, in the quantum scenario,

prohibition: physical theoiy cannot give meaning to something

of representing a reality to which measurement would provide

it defines as escaping prediction. In other words, the strength

access while remaining autonomous from it simply reflects


the fact that physics has finally acknowledged epistemological norms. Another thesis, associated with quantum theoiy but
extendable elsewhere, "explains" the limits of measurement in
the quantum scenario, which makes no sense from the positivist
point of view. This thesis states that the interaction needed for
measurement "perturbs the phenomenon." It would then be the
perturbation associated with the operation of measurement that
would create the necessity of choosing what to measure. Here,
the choice is between those variables that are being measured
and those that will be perturbed as a result of the operation. Correlatively, the preparation of a system that allows physicists to
predict a physical quantity with certainty, one corresponding to
position, for instance, will have as its price a radical uncertainty
concerning some other quantity, here the quantity of motion.
The perturbation, then, would provide the key to Heisenberg's

32

of the argument lies in the way it operates. First it ensnares


physicists by convincing them that a common view of reality can be shared, then it strikes by confronting them with the
prohibition. The ongoing discussion about "hidden variables"
reflects the strength of the argument, which assigns a crucial
role to the rules that scientific knowledge should or should not
obey. Because the measurement of certain variables results
in the impossibility of assigning a well-defined value to other
variables, the values of the latter are "hidden." Do we have the
"right" to refer to something to which we don't have access, to
claim that, prior to the measurement that will perturb them,
those variables have a value, even if it remains hidden to us? Or
should we eliminate all reference to hidden variables as being
devoid of meaning? Can we use Einstein as an example, when
he claimed that the simultaneity of two distant events must be
recognized as devoid of meaning?

34

NIELS

BOHR'S

LESSON

NIELS

BOHR's

LESSON

35

Bohr, however, perceived the obligations that follow from

In influenced by this relationship of resemblance. For Bohr,

the new quantum mechanics quite differently. The first quantum

1 lie equation that connects the evolution of the wave function

model of the atom, of which he was the author, unhesitatingly

wi I h the Hamiltonian of the "quantum system" (the Schrodinger

connected contradictoiy physical elements. Classical mechan-

equation) has nothing to do physically with Hamiltonian phys-

ics, more specifically, electrodynamics, was both required and

ICH, The equation does not imply that what is described is deter-

denied by the two postulates of the model: electrons were in a

mined in terms of position and velocity, that those variables are

stationary orbit around the nucleus, whereas they should have,

"objective" in the sense that they can be attributed to an object.

like eveiy electrically charged body in motion, created a field,

In quantum mechanics, Bohr insisted, the data about position,

that is, gradually dissipated their energy. The amount of light

velocity, and all other measurable quantities, cannot be attrib-

(now, the photon) emitted or absorbed by an atom reflects the

uted to any reality. It is their formal presentation that makes

individual, nondecomposable, nature of the event constituted by

l hern variables, but their physical meaning relates them to the

an electron's transition between two stationary states. In other

detection devices used to measure them. Those devices create

words, Bohr was convinced that the quantum atom constituted

I he possibility of making an observation about a quantum being,

a new physical reality, requiring a new physics, releasing the

but what is observable cannot be attributed to this being while

physicist from any obligation toward classical mechanics. Bohr

ignoring the device. In other words, for Bohr, measurement of

affirmed the "acausal" nature of quantum physics, a statement

I he quantity of motion has no effect on position whatsoever. It

that should be understood in a strictly technical sense: quantum

reflects the intervention of a detection device that actualizes the

reality cannot be understood in terms of the causal measure-

"quantity of motion" observable.

ment on which the classical edifice had been built.


And yet, in the second formulation of quantum mechanics

Bohr retrospectively defined classical dynamics, from Galileo to Hamilton, as determined by an idealization that becomes

of 1927, the two variables of position and motion, which in the

misleading whenever the physicist addresses the quantum

classical description exhaustively determine dynamic behavior,

world. What did causal measurement rely on? As we saw in Book

retain their central position; and energy, or the Hamiltonian

II, it was causal measurement that enabled Galileo to create the

operator, which articulates those variables in classical mechan-

very concept of instantaneous velocity and, with the rational

ics, continues to do so, although differently. Quantum formal-

mechanics of the eighteenth century, gave physicists the abil-

ism, centered on the wave function, is not, like Bohr's model,

ity to measure force. In both cases, the visibility of the body,

a "description" that, at the cost of two scandalous postulates,

the possibility of determining where it is, is presupposed. For

designates the atom as an actor and allows us to indicate which

Galileo especially, the equal sign implies that we know both the

"aspect" of that actor's activity is measured through detection of

initial height and velocity of the moving object: the height that

the photons it absorbs or emits. The formalism, which intro-

characterizes it at a later moment will then allow us to deter-

duces an equation that affirms the relevance of the categories

mine the amount of velocity it has gained or lost since leaving its

of Hamiltonian dynamics, is a postulate in every sense of the

point of departure, and thus its resulting instantaneous velocity.

word. I will return to this later. Here, I want to emphasize the

The idealization arises because the possibility of spatial deter-

uniqueness of Bohr's position, for he never allowed himself to

mination is taken for granted, it can be "seen," and this visibility

37

NIELS

BOHR's

LESSON

N r E L S BOHR's

LESSON

45

hides the energy interaction it presupposes. Even in astronomy,

followed from the project of interrogating quantum beings in

we know the Sun's position only through the emission of light.

terms that allow us to interpret detection, to claim that if a mark

In both cases the idealization that overlooks this interaction is

IN produced on a photographic plate, the quantum being was

legitimate in that the order of magnitude of the energy interac-

"I here," or that if electrons reflected by a target lose energy, it is

tion required (given by Planck's constant) is negligible com-

because they were the "cause" of a recoil of the target particles,

pared to the energy associated with the motion of the body being

which received energy equivalent to the energy lost by the elec-

characterized. Detection has indeed taken place, but it has taken

I i ons. In other words, just as the laws of classical electrodynam-

place at the level of the retina or the photographic plate. Who

ics were required by Bohr's model of the atom, which happened

would dare to claim that the Sun is in any way affected by the fact

to contradict them, the classical idealization, made untenable by

that the photons it emits enable us to locate its position? Doesn't

(lie finite interaction any detection entails, must still be main-

it emit them regardless? Irresistibly, we think: it is where it is.

tained for the interaction to become a measurement, productive

It is this irresistible conclusion that quantum mechanics,

of meaningful information.

according to Bohr, obligates us to resist. Neither position nor

The great contrast between perturbation and Heisenberg's

quantity of motion answers to the idealization of mere location.

uncertainties, on the one hand, and Bohr's production of deter-

To determine the position of a particle is to answer a question we

minateness, on the other, is related to the respective status

have taken the means to make decidable by the appropriate use

each confers upon classical mechanics. From the point of view

of a detection device. And in order to be answered, the respective

inspired by Heisenberg, it appears to be something of a lost par-

questions about the localization and evaluation of the quantity of

adise. The fact that, in the quantum case, the perturbation can-

motion imply logically incompatible means. The term "logic" is

not be neglected (Planck's constant, h, ensures that it will have

important, for it emphasizes that it is information, intelligible

an order of magnitude comparable to what is being measured)

information, leading to reasoning, that we demand of our detec-

can only inspire a certain nostalgia given that this perturbation

tion devices. Thus, the device that provides information must

has only negative value, creating a kind of slow torment for the

also produce meaning. And it is what is implied by the produc-

physicist. The Bohr interpretation, on the other hand, requires

tion not of data as such but of meaningful information, infor-

a crucial distinction that mechanics, ever since Galileo, has,

mation about position, that is incompatible with the production

according to him, spared the physicist. Causal measurement

of meaningful information about velocity. The first requires a

authorized description and reason to coincide. Between the

rigid device, which makes energy exchange uncontrollable; the

object defined by measurement and the "measured reality" there

second must introduce causal measurement, measurement of

was thus no interruption requiring that an observer be impli-

the transfer of energy, which prevents us from rigidly attaching

cated or introduced, or obligating the physicist to recall that

the device and makes position uncontrollable.


Bohr insisted that our measurement devices are part of classical mechanics by definition, and he was often heard expressing what sounded like an arbitrary limit to the ingeniousness
of future devices. However, he adhered to the obligations that

measurement had to be understood in terms of intervention,


mediation, or the production of meaning. Reality was determined "in itself" by the very mode of determination introduced
by measurement. For Bohr, the lesson of quantum mechanics
was not so much about knowledge as it was about reality itself,

39

NIELS

BOHR's

LESSON

more specifically, the fact that reality did not, either in classi-

N r E L S BOHR's

LESSON

45

I lie spontaneous return of an excited atom to its fundamental

cal mechanics or in quantum mechanics, dictate its categories

'U.iliv'The physicist would be free to believe that another level of

to the physicist. Variables, whether classical or quantum, refer

ih'Kcription exists, one that "explains" probabilistic laws the way

observables to variables, the meaning of which is the function

lour-dimensional mathematical reality "explains" our obser-

of which they are variables, but they never authorize the physi-

vnlions in space and time. Einstein's opposition to quantum

cist to speak of a "functional" reality. They merely express the

mechanics had less to do with a conflict between deterministic

fact that the physicist has taken the means to characterize real-

and probabilistic visions of the world than with the question of

ity as a physical-mathematical

the relationship between "reality" and "determination."

function. For Bohr, with regard

to the requirements of a functional definition, "reality" has to

In 1935 Einstein launched a final attack on the idea that

be referred to as indeterminate, independently of the measur-

quantum mechanics constitutes a "complete theory." He coau-

ing device able to provide a determinate interpretation to an

I bored an article with Boris Podolsky and Nathan Rosen titled

observation.

"Can Quantum Mechanical Description of Physical Reality Be

Here, it is appropriate to mention the radical distinction

Considered Complete?" in which the authors (customarily

made by Gilles Deleuze between the virtual and its actualiza-

referred to as EPR) defined the obligation to which any theory

tion, on the one hand, and the possible and its realization, on

claiming to be complete must comply. They write: "Every ele-

the other. The only thing "missing" from the possible is exis-

ment of physical reality must have a counterpart in physical

tence. Indeed, this is what is presupposed by the measurement

theory." The element of physical reality in turn had to satisfy a

device assumed by dynamics: a body can have any possible posi-

requirement: "If, without in any way disturbing a system, we

tion or velocity; measurement occasions the transition from the

can predict with certainty (i.e., with probability equal to unity)

possible to the real of one of those values. Actualization, on the

the value of a physical quantity, then there exists an element of

other hand, is associated with creation. It implies a change in

reality corresponding to that quantity." Commenting on their

kind, not the determination of a preexisting possible. "The vir-

definition, the authors remarked that it was in agreement not

tual possesses the reality of a task to be performed or a prob-

only with classical mechanics but also with the constraints that

lem to be solved: it is the problem which orientates, conditions

quantum mechanics claims to adhere to. In other words, EPR

and engenders solutions, but these do not resemble the condi-

do not provide a general argument against quantum mechan-

tions of the problem." In reply to the concept of the problem in

ics. They offer to construct a critical argument that will find fault

Deleuze we have Bohr's concept of choice. For Bohr, the choice

with it on the veiy grounds it has defined for itself. They intend

of a detection device is the choice of the problem by which the

to show that there exist quantum "elements of reality" that have

production of determination (or the actualization of the virtual)

no counterpart in theoiy.

assumes meaning.

The expression "without in any way disturbing a system"

We are familiar with Einstein's well-known complaint,

indicates that the EPR approach requires Heisenberg's per-

"God does not play dice with the universe." But Einstein, like

turbation interpretation. It will involve finding a way around

all the critics of the Copenhagen group, would have been vastly

Heisenberg's prohibition regarding hidden variables. The EPR

more satisfied if quantum mechanics had in fact presented us

article introduces a thought experiment in which a "quantum

with a probabilistic description, like that of nuclear decay or

object" spontaneously breaks down into two subsystemslet's

40

NIELS

BOHR's

LESSON

call them A and Beach of which, based on what the experi-

N r ELS

BOHR's

LESSON

45

mines down to knowing if we can state that they are separated

menter decides, could be subject to measurement. The crucial

in any unambiguous sense. It is not enough to say that measur-

point is that, because of the unique object that produces A and

ing one does not affect the other. We must also be able to show

B, we have information about A and B such that they preexist the

I hat it does not affect the possibility of measuring the other. And

measurement they may be subject to. For example, if the single

II is here that Bohr strikes back. If we can use the information

system was a molecule with spin o, we "know" that the spin value

derived from the single system from which the two subsystems

of one of the "particles" must be +1 and the spin of the other - 1 .

arise, we do so to the extent that these two subsystems are perti-

Therefore, if we measure the spin of A in a known direction,

nently defined by their common origin, that is to say, are related

let's call it x, we can, without disturbing B in any way, deduce the

lo (he same experimental device through which the information

value of its spin in that direction. What prevents us, then, from

about the original system has acquired its meaning. Therefore,

measuring on subsystem B another, complementary observable

no matter how far apart the two subsystems are, they continue

(here, the spin value in directiony, for example), an observable

lo form, from the point of view of the unambiguous formulation

that quantum mechanics states can be given no physical mean-

of the information we have about them, a single system. And in

ing if we give physical meaning to the first. Naturally, this mea-

I his sense, we cannot say that measurement of one of the two

surement will make it impossible to actually measure the spin

particles allows us to characterize the second while leaving the

of B in the x direction, but this is, in any case, redundant as the

physicist free to make another measurement of this second par-

value can be determined by measurement of A. And reciprocally,

ticle. In fact, the measurement results in the "destruction" of the

measurement of the spin value of B in the / direction allows us

single system, in the sense that its singleness could serve as a

to assign to A the value of the observable that the previous mea-

common reference, leading to a determinate inference, for the

surement of A had made inaccessible. Therefore, each particle

two "separate" particles. If measurement can be used to assign a

is characterized in itself by two variables that quantum mechan-

spin to what then becomes the "second" particle, this is another

ics states cannot have physical meaning simultaneously. Conse-

way of saying that this second particle has been subject to mea-

quently, quantum mechanics is not complete.

surement: the "first" particle has become an integral part of a

Leon Rosenfeld recounted Bohr's perplexity when confronted with this objection.4. His perplexity arose not from
the difficulty of the response but because the EPR experiment
presented no problem to him at all. On the contrary, for Bohr
it served as an excellent restatement of his own message, a
message he himself was unable to contemplate without a cer-

measurement device that detects the spin of the second. And


the physicist's choice of a device is irreversible; it actualizes this
"second particle" to the extent that it is separated and characterized by the measurement performed. A similar measurement
made directly on this particle would confirm the value of the
observable in question with a probability of one.

tain uneasiness because it could not be reduced to the idea that

In other words, for Bohr the "spatial" separation had no

measurement "disturbs" the measured system. In the EPR case,

physical meaning in itself, for to the extent that their common

he stated in his response that it is assumed that at the moment

origin had to preserve a precise, unambiguous meaning, the

their spin is measured, the two particles are arbitrarily dis-

two subsystems are in fact not separate from the point of view

tant from one another and do not interact. The whole question

of an observer. Measurement of the one actualizes the physical

43 N I E L S

BOHR's

LESSON

meaning that will be given to their shared origin, allowing us to

N r ELS

BOHR's

LESSON

45

demand that reality be theoretically defined as independent.

infer the result of a measurement of the other; and it is in this

EPR continued to require that each "element of physical reality"

unambiguous sense that it can be considered measurement of

authorize this formidable claim: the possibility of defining that

the other subsystem. But this actualization destroys any possi-

element, of assigning to it a physical and theoretical meaning

bility of once again referring to this common origin for other,

independently of any practical connection.

complementaiy measurements. Taking "another" measurement

It is important to understand that the requirement EPR

of the second subsystem would, as always, amount to introduc-

reproached "quantum reality" (as defined by quantum theoiy)

ing a device that destroys the meaning of the first measurement

for failing to satisfy has nothing to do with what other practices

and creates another actualization, logically contradicting the

require of their "reality." Even less with the knowledge one

first.
Once again, Bohr used the consistency of formalism to defeat

human can have of another. How can I even dream of knowing


the other "in itself," independent of the relation I have with it,

those who thought they had detected a weakness in the theory. If

independent of our respective abilities to form relationships

EPR were right, quantum mechanics would not be incomplete

with one another and with others? The notion that knowledge, in

but wrong. And indeed this was what was at stake when the

this instance, is dependent on words, on contexts of meaning in

implications of the formalism in a more recent, EPR-like case

which those who know one another participate, is not something

were challenged experimentally. As a result, the hypothesis of

to regret. It is, rather, the possibility of knowing in the absence

the kind of "hidden variables" postulated by EPR, which are said

of a relationship that is a nightmare. Should we regret what takes

to characterize the quantum system in itself, in spite of Heisen-

place in a laboratory, where phenomena are effectively staged,

berg's prohibition, has since been refuted. 5 But Bohr's answer

purified in such a way that they become experimentally mean-

was not vindicated. It is still seen as confirming critics' darkest

ingful, acquiring the power to authenticate their representa-

suspicions; Bohr "denies" reality, he "prevents" us from refer-

tion? No one would dream of imagining that the necessity of a

ring to what we "know."

laboratory, of the devices that are used to transform an "empiri-

Obviously, Bohr never denied that reality "exists," or that

cal" fact, subject to a thousand and one interpretations, into

it is "knowable." Reality is clearly "knowable," as the existence

an "experimental" fact, implies the "unknowable" character of

of quantum mechanics demonstrates. But being "knowable" is

reality. Quite the contrary, experimenters are all the more "real-

something quite different than the possibility of knowledge that

ist" to the extent that their practice obligates them to fiercely

critics demand. What they express is nostalgia for that blissful

distinguish between "fact" and "artifact," that is, to distinguish

situation where reality itself seems to dictate the categories of

between those cases where "reality" has indeed satisfied the

its definition. For the strategy adopted by EPRinitially claim-

requirements that define it as a reliable witness and those where

ing that quantum formalism accepts the obligations of a theoiy

the device has extended the power of interpretation to produce

that claims to be complete, and then revealing a situation where

a "false witness" who cannot but confirm that reality. Labora-

it betrays those obligationscontinued to define a "reality ele-

tory practice connects "reality" not to the possibility of predict-

ment" (which had to have a counterpart in theoiy) in a way that

ing without intervening, but to the possibility of an interaction

confused the demand for an "independent reality" with the

productive of evidence whose meaning can be determined.

NIELS

44

BOHR's

N r E L S BOHR's

LESSON

From this perspective, Bohr's indeterminacy does not signify

LESSON

45

ciil I cagues may seem profoundly satisfying and entirely healthy.

unknowability. It reminds us that every determination is pro-

I I lose who treat the loss of this dream as something dramatic

ductive of a link that carries meaning, creates the ability to make

lor Ihought and assimilate it to the loss of an independent real-

a difference for the one who has the means to determine.


Bohr, therefore, did not give up trying to "know" reality. He

ily, intelligible and describable, thereby translate the polemical


definition of "physical" reality to which they adhere, a defini-

remembered that it was only because the bodies interrogated by

tion that pits this reality against all the "realities" presented by

classical mechanics allow themselves to be presented in terms of

other practices. 7 Correlatively, the way in which Bohr general-

the idealization of mere location that, in Galileo's lab and that of

ized the obligations entailed for him by quantum mechanics

his heirs, experimental practice appeared to have simply staged

i lie idea of "complementarity"isn't so much fascinating as it

a reality determined in itself and by itself. The reality he wanted

is relevant for other practices. For the first time, physics is not

to reject was not the one presupposed by experimental practice,

presented as a model but as something subject to an obligation,

or the one each of us presupposes when addressing the world

one that can be transposed to other sciences, yes, but as a form

and other humans. It was the reality of the Queen of Heaven,

of repetition rather than something to be complied with. What

the dream of a reality whose truth could be attained indepen-

is at. stake is the singularity of each individual science whenever

dently of any practice, any question, any relationship. It was the

the complementary affirmation makes itself heard: there is no

reality that had been preserved by Einstein's general relativity,

answer without a question. In other words, each science must

to the extent that the different points of view situated in space

undergo, in its own way, the challenge of statements such as: no

and time, and creating an observable phenomenon in space and

knowledge content can win its independence from the question

time, could be deduced from four-dimensional mathematical

that gives it meaning; no question can gain its autonomy from

truth. It was the reality that everyone who had tried to create a

the choice from which it proceeds; no choice can prevent its

quantum mechanical beyond wished to reestablish. Even though

selective nature from being taken into consideration, can ignore

that beyond might assume the seductive and quasi-mystical

what is excluded from being presented so that what is chosen

appearance given to it by a physicist like David Bohm, that of an

can present itself. 8

"undivided, multidimensional, and implicated" totality, based


on which we can understand the engendering, or the unfolding
of our "explicate order" as secondary, limited, and valid only in
certain contexts. 6
In Book I of Cosmopolitics, I addressed the problem of the

Nonetheless, the fact is that physicists have not abandoned


the dream. We can even say that high-energy physics, when it
addresses the mathematical symmetries that characterize its
objects rather than behaviors in space-time, has reinvented
the dream, which then becomes, as Heisenberg noted, frankly

"vocation" of physicists, which commits them to disqualify-

Pythagorean. Physicists no longer require of the interrogated

ing all other knowledge practices. The "vision of the world" to

reality that it subject itself "in itself" to the determinations in

which physicists' faith is addressed situates them in an "else-

terms of which we measure it. They address symmetry proper-

where" from whose vantage point all other practices are judged

ties that, independently of measurement, characterize the math-

by their benefits, limitations, and misinterpretation. From this

ematical beings presented by theory. In this way they take as

point of view, the "end of the dream" to which Bohr exhorted his

their object the symmetry properties of the wave function even

4,6

NIELS

BOHR'S

LESSON

NIELS

BOHR'S

LESSON

4-7

though the reduction of this wave function, which gives mean-

interesting reason for the intellectual dissatisfaction of physi-

ing to observable properties, destroys those symmetries. It

cists with the Copenhagen interpretation, we must set aside

is important to note that the separation thereby instituted

commonly used slogans such as realism, determinism, posi-

between "objective" symmetry and determination by measure-

livism. And we must reformulate the "dream" that physicists

ment in itself constitutes the negation of the so-called orthodox

refused to abandon in more technical terms, using terms that

interpretation. For Bohr, the wave function had no meaning at

singularize their passion and do not condemn it a priori.

all independent of its "reduction." The very idea of "posing the


problem of the reduction of the wave function," that is, considering the transition from probability amplitudes to determinate
probabilities as a problem, was devoid of meaning for him. The
fact alone that it was in these terms that the "problem of measurement in quantum mechanics" is ordinarily presented accurately reflects the fact that physicists have not felt in the least bit
obligated by the obligations that, for Bohr, follow from the novelty of quantum mechanics.
It is impossible to acknowledge that Bohr was right and
denounce those who didn't follow him without transforming my
approach into a critical, and thus normative, endeavor. Bohr,
like Duhem, failed. But we can use his failure to better understand the practice to which his proposition was addressed.
"Bohr was incomprehensible, his language was obscure, he
would think out loud." I don't want to let such statements stand
in my way, even though they might be relevant. Not only did
people listen to Bohr, intensely, not only did he continuously
struggle to express his thought in ever more lucid terms, but he
had, in the person of Leon Rosenfeld, a faithful and perfectly
intelligible interpreter. Anecdote alone is inadequate. Here,
misunderstanding must primarily be understood as a refusal to
understand, that is, a refusal to consent to understand, as William James would put it. And psychological interpretation is of
no more help than anecdotal evidence. It seems to me that the
refusal to consent has not been adequately interpreted in terms
of physicists' attachment to their dreams of determinism, power,
and omniscience. In order to have a chance of discovering an

QUANTUM

IRONY

49

I irrn reserved for use by astronomers, was transposed to Bohr's


quantum atom, as if it formed a natural framework in which to

formulate the rules of quantification for the series of "orbits"


that defined the energy transitions introduced by Bohr to inter-

Quantum Irony

pret atomic spectra.


The irony here consists in the sudden encounter between
I lie "obscurity" of the atom's acausal behavior and the most
luminous form of intelligibility. What is obscure, in the sense
of being postulated with no explanatory justification, is the idea
of a "quantum stationaiy state," where a charged body such as an
electron is able to remain indefinitely in orbit, and of a "quanlum jump" that occurs instantaneously, without any "transition" between two states. Its luminous intelligibility is found in

From now on, the question is no longer one of interpreting

the cyclic Hamiltonian representation. As mentioned earlier,

quantum formalism but of the problem associated with the for-

it is the only Hamiltonian representation that does not present

malism itselfwhat it commits physicists to and what it prohib-

physicists with a problem to be solvedthe integration of the

its them from doing. I would like to show that it's not impossible

Hamiltonian equations; rather, it presents the problem in terms

that the intellectual dissatisfaction of physicists reflects the

that coincide with those of its solution.

irony with which quantum novelty was inscribed within the tradition of physics.

In other words, whereas standard dynamics, with its interactions, accelerations, positions, and velocities, was equipped

This irony began early on, with Bohr's first model of the

for writing differential equations whose integrationthat is,

atom. As we have seen, in order to incorporate the well-defined

(he effective description of the spatiotemporal behavior of the

spectral lines that served as the signature of each chemical ele-

systemhad been perceived as a kind of Holy Grail, the atom

ment in the periodic table, Bohr had imagined the electrons to

appeared to require the analog of a cyclical representation, syn-

be distributed in stationary orbits of well-defined energy. But

onymous with the Grail's possession. The atom's stationaiy states

we have already encountered a version of the notion of a well-

correspond to the way a dynamic system can be represented if

defined stationaiy energy orbit in Book II. The Hamiltonian

and only if the problem of integration has been solved: in terms

representation resulted in such "stationary movements" when-

of "free" modes that are independent of one another. The atom,

ever it made use of cyclical variables. In this case, each degree of

which had entered physics as an actor in random kinetic events,

freedom is characterized by an action variable, J, independent

therefore creates, where its behavior is most obscure, the pos-

over time, the system being represented as a collection of free

sibility of a junction with the most elaborate, the most refined,

points that do not interact with one another. Through the work

the most singular fiction produced by dynamics: a fictional

of the mathematician Arnold Sommerfeld and the astronomer

description that absorbs interactions, and the accelerations

Karl Schwarzschild, this representation, which had until then

they determine, into the definition of the system's constituents,

48

6aQ U A N T U M

IRONY

QUANTUM

I R O N Y 30

so that each of them reciprocally constitutes a local expression

between Hamiltonian dynamics and the atom, a reciprocal

of the overall identity of the system. And correlatively, this fic-

capture that short-circuits the question of the resolution of the

tion acquired the means to claim a kind of privileged relevance

dynamic problem and asserts its solution, the stationary state

because it is immediately confirmed by the energy spectrum

being described in the language of integrable dynamic systems.

that identifies the individual atom.

As for the quantum jump, it associated the atom with the active

Where is fiction? Where is reality? In classical dynamics,

protagonists of kinetics, those whose activity had to be detected,

each stationary state is represented as "alone in the world," its

those whose "initiative" had to be recognizedwhen will it emit

definition expressing the totality of a system itself defined as

;i photon? Between the two concepts, no hierarchy should be

closed, interacting with no environment. Accordingly, the equa-

possible. It is on the basis of the energy differences between the

tions defining the postulated quantum stationary state prohibit

levels, that is, the transitions, that stationary states can be char-

us from defining the atom as a system interacting with an envi-

acterized. Logic seems to demand that accepting the effective

ronment. This implies that the atom defined by its stationary

symmetry between the two postulates means challenging the

states is literally unobseivable since there is no observation with-

previously accepted claim that kinetic events correspond to an

out interaction. But if we are, nonetheless, able to know some-

incomplete dynamic description (see Cosmopolitics, Book III).

thing about the atom, it is that the cyclical representation, in the

But logic does not rule the history of physics.

case of the Bohr atom, does not tell us everything. The quan-

The irony of the situation derives from its duality, from the

tum atom does indeed "interact" with the world; it emits and

two different ways it can be described. On the one hand, we

absorbs light quanta. In 1916 Einstein showed that the "transi-

could say that Maxwell's "Queen of Heaven," whose triumph we

tions," the "quantum jumps," if they were to satisfy the laws of

followed in Book III, had finally touched ground, where its rel-

Planck's black-body radiation, had to correspond to two types of

evance could only be confirmed by stating, simultaneously and

processes. Either the transition (the emission or absorption of a

symmetrically, the relevance of what escaped its power. On the

quantum of light) is caused by an external electromagnetic field,

other hand, we could sayand this is the path de Broglie and

or it is spontaneousan excited atom spontaneously "falls back"

Schrodinger followedthat the only pocket of resistance to

to a state of lower energy while emitting a quantum of light. In

dynamic intelligibility was now successfully identified: if they

other words, the "excited" stationary states have a lifetime, like

were able to "smooth" the quantum jumps, Hamiltonian har-

radioactive nuclei. Whereas Bohr's first postulate creates a con-

mony would reign over the physical world.

nection with the dynamic representation of a world where nothing happens, where nothing can happen, the second, that of the
"quantum jump," refers us to kinetics, the science of events.

De Broglie and Schrodinger did not succeed, but their ambition still haunts the formalism surrounding the "wave function"
defined by the Schrodinger equation, tempting physicists to

The two concepts postulated by the Bohr atomthe sta-

revive it. This formalism indeed breaks the symmetry of Bohr's

tionary state and the quantum jumpwere in a relationship of

two postulates. The definition of stationary states corresponds

reciprocal presupposition. However, they are very differently

to a solution of the Schrodinger equation, but the discontinuous

inscribed in the mathematical physics of the time. The station-

quantum jump no longer corresponds to an "event" that would

ary state resulted in the most astonishing reciprocal capture

"happen to the atom." It is the "reduction of the wave function,"

QUANTUM

QUANTUM IRONY

IRONY

the transition from probability amplitudes to probabilities,


that takes the place of the "quantum jump," because, "like it,"
this transition is associated with the possibility of assigning
to a "quantum system" a determinate value of an observable
property.
Here we encounter the greater irony, superimposing itself
on the first and stabilizing its effects. A form of mathematical
wordplay has literally caused the kinetic actor to disappear, to be

,1 period of life on Earth when there were no humans, much less


I ihysicists, but a period, nonetheless, when the world that quantum mechanics claims to describe resembled our own. Vigier
und Rosenfeld knew well, given that they were Marxists, that
behind this exchange loomed the menacing shadow of Lenin,
who, in Materialism and Empirio-Criticism, denounced as idealis! and solipsistic those who might be suspected of claiming that
"reality" depended on their perception. 1

replaced by the question of the reduction of the wave function


associated with "measurement." Bohr's atom was "knowable"
only because it emitted and absorbed light, but the absorption or emission event, the condition of measurement, was
not defined as a measurement. If we live in a world of color, it
is because, according to the terms of Bohr's model, atoms con-

53

ROSENFELD:

All that we can say about those Saurians is based

on what we can see of them . . . I cannot see the least philosophical difference between descriptions of the state of the
world in the secondary epoch and the description of the
world as it is t o d a y . . .

tinuously absorb and emit photons. The fate of the majority of


I suggest that if Professor Rosenfeld does maintain

photons remains undetermined. Some of them hit our retinas,

AYER: . . .

others strike a photographic plate in a physicist's laboratory and

this position he must do it in a form that would escape Pro-

result in a measurement, but the majority of them participate in

fessor Vigier's trouble about the Saurians. He must say that

other interactive adventures that we can only imagine. Yet, after

in talking about the Saurians he is talking not about any-

1937 this "realist" language could no longer be used. The emis-

thing he did observe but about something somebody might

sion and absorption probabilities have physical meaning only

have observed, had he been there, even if he wasn't there. To

in terms of the measurement device and, one way or another, the

make this theory work, you have got to do it in terms of the

device appears as "responsible" for the transition from prob-

possibility of making the relevant observations and not in

ability amplitudes to probabilities. Regarding the atoms that

terms of actual observations . . .

were "free with respect to the physicist," the ones responsible


for the bright colors of flowers that attract butterflies, nothing

ROSENFELD:

I am glad that you have mentioned thatthis

could be said. Spontaneous emission was no longer a property

is exactly what I meant. But I didn't mention it explicitly

of the atom, through which it could become known; it is the pro-

because I thought it was quite obvious.

duction of knowledge about the atom, the record of a photon of


carefully determined frequency, that affects the representation
of the atom, that is, reduces the Schrddinger wave function.
Where I have referred to colored flowers and butterflies,
Leon Rosenfeld, David Bohm, and Jean-Pierre Vigier confront
the question of dinosaurs, that is, the possibility of "knowing"

VIGIER:

I was veiy happy about what Professor Ayer said and

I also agree with Professor Rosenfeld, for this is the first


time I have heard Professor Rosenfeld disagree with a plain
positivistic sentence. I would put the thing in a stronger
form: I do not think that things which exist are things which

QUANTUM IRONY

QUANTUM IRONY

55

might have been observed. This is where the split comes in a

Hosenfeld. When Rosenfeld states that "things" exist, whether

very clear form.

or not they are observed, and that quantum laws apply whether
or not there is an observer, Vigier thinks he has wonquantum

ROSENFELD: I don't say that. Don't continue on that line

laws are, therefore, laws of "nature." But then Rosenfeld says no,

because I do not say that things only exist in so far as they

lo Vigier's horror. If Rosenfeld answers "no," it is because Vigier

could have been observed. All the statements we make about

lias failed to take into account that the laws of quantum mechan-

the world are necessarily descriptions of a state of affairs,

ics, like every "objective" statement, apply only to "states of

of mind, of material, that an observer might perceive if he

affairs," not to beings. And a state of affairs must include refer-

were placed in those particular circumstances.

ence to a well-defined possibility of stating the affair. The colored flower is a meaningful affair for the butterfly. And when we

VIGIER: Let us say then we agree that the world exists

apply quantum mechanical laws to it, it no longer refers to but-

outside any observer. Did the laws of quantum mechanics

terflies but to the photographic plates that allow us to describe

apply to the world at a time when there were no observers

its atoms in terms of observable quantum facts.

present?

To speak of irony is to claim a relative sympathy for the victims of irony. If paleontologists can reconstitute a dinosaur

ROSENFELD: Of course.

from fossilized remains, if molecular biologists are not troubled


when they speak of the DNA molecule as if it existed or acted,

VIGIER: OK. If you say then that the laws of quantum

why should physicists be forbidden from referring to what they

mechanics did apply at that time, then the laws of quantum

investigate as an existent? Should they really say good-bye to

mechanics are real, objective, statistical laws of nature,

any possibility of constructing the representation of an actor to

which have nothing to do with the observer, and are verified

which we could "really" assign responsibility for our measure-

whether there are observers or not.

ments of it?
This is the intellectual rebellion that Erwin Schrodinger

ROSENFELD: NO.

introduced with his famous parable of the "cat" in the box. A


cat is enclosed in a box together with a device that triggers the

VIGIER: YOU

can't change your position and say something

two minutes ago and another thing n o w . . .

release of a poison gas when a radioactive particle decays. This


means, Schrodinger emphasizes, that the cat "actually" dies
when the decay event occurs. However, quantum mechanics

Why quote this exchange here, when I've already indicated

does not allow the cat to die until the box is opened and the cat's

that it was no longer a question of "interpretation"? Because it

statedead or aliveis observed. The cat is, indeed, a measur-

exposes what is most often dissimulated in matters of inter-

ing device like any other, and quantum mechanics demands

pretation: a genuine malaise when confronting the role that

that we accept that the particle will never "really" disintegrate,

formalism gives to observation. Vigier's frustration is caused

any more than an excited atom will fall back to its fundamental

by the technically, rather than deliberately, ironic position of

state while releasing a photon, unless measurement has taken

QUANTUM

IRONY

place. As a consequence, we must conclude that the "event,"

QUANTUM

IRONY

57

nl ilie severe dissatisfaction aroused by quantum formalism is

here a chain of events ultimately leading to the cat's death, can

mi longer the impossibility of "revealing" a real that, like the

be postponed indefinitely "until" the conditions of observation

Queen of Heaven, would have the power to equate description

have been fulfilled. This means that "until" that time, until the

.mil reason or, like the four-dimensional mathematical reality

measurement takes place, the cat, like the radioactive particle,

nl general relativity, the power to designate a truth that is "supe-

will be described as a superposition of probability amplitudes,

i tor" to our measurements, that gives them their meaning. The

in other words, as being both "dead and alive" rather than being

Lid l hat reality does not satisfy this requirement is not the prob-

either "dead" or "alive" with definite probabilities.

Inn. The fact that light does not allow us to construct an identity

Schrodinger's thought experiment has since been assimi-

that transcends measurement and explains its "phenomenal-

lated and presented as a scenario in which quantum mechanics

ization," either as a particle or as a wave, is not the problem.

is assigned the fascinating power to be able to claim that the cat

Hul why does giving up the requirement of a transparent reality,

is indissolubly "dead-alive." In fact, it should instead be thought

intelligible in itself, why does acknowledging complementarity

of as a cry of rebellion, a cry of passion. Quantum mechanics is

,iik1 the fact that our descriptions include an irreducible refer-

not complete and cannot be considered complete! Schrodinger

ence to measurement, lead us toward a language whose syntax

was unable to define just "what" quantum mechanics was miss-

makes measurement the only thing responsible for the world's

ing, but to him the Copenhagen position was perverse. We

observability? Cats exist, they observe and die! In the world,

"know very well" that the nucleus decays on its own. We know

meaning is created "without us," and it is created without ref-

that the cat dies. We could say that these are "actors" to whom

erence to measurement. The butterfly is guided by the photons

things happen. On behalf of what can we consider as satisfactoiy

emitted by flowers. Why must physicists agree to be trapped in

a formalism that makes physical measurement the only "actor"?

t he sudden, stifling abstraction of criteria of meaning that refer

Of course, measurement alone provides the information that

to measurement alone? Why must they accept a definition of

has meaning for physicists. But why should they give up invent-

I heir practice that obligates them to cut themselves off from the

ing the theoretical syntax that would "cause" the disintegrat-

world, imprison them in meanings they, and their measuring

ing nucleus, or the excited atom falling back to its fundamental

devices, bring into existence?

state,to "exist"?

It is the significance of the place known as a "laboratory"

Vigier and Lenin were wrong. The question is not one of

that is in question here. As Bruno Latour, in discussing Pasteur,

idealistic solipsism, which questions the existence of a real-

has admirably shown, what is made in a laboratory never really

ity "without us." It is, rather, as Schrodinger's protest vividly

leaves the laboratory. Rather, it is the laboratories that prolif-

expressed, a question of the radical divorce instituted by quan-

erate, and they do so in placeshospitals, the food-processing

tum formalism between reality without us and reality "for us."

industries, government agencies, in the case of Pasteurwhere

It is a question of the genuine "claustrophobia" of the physicist,

the scientific proposition is to become relevant. Wherever a

who is suddenly obligated by formalism to no longer refer to

science appears to have "spread," we find that it is the devices,

the signifying "counterpart" of practices that are productive of

practices, the gestures of sampling and measurement, the

meaning. And the term "refer to" is important here. The source

rules of interpretation that have been successfully adapted and

6aQ U A N T U M

QUANTUM

IRONY

implanted. Yet, the laboratory of Pasteur's heirs has continued to


be transformed with the appearance of new instruments as well

IRONY

59

To describe what quantum irony calls impossibility within H getting wrapped up in the question of a "reality knowable in

as new problems associated with new circumstances. The labo-

IIHCII'," of "strong objectivity," the concept of delegation intro-

ratory certainly creates devices that are both filter and ampli-

duced by Bruno Latour is extremely useful, for it integrates,

fier, purifier and stabilizer, but what it detects and cultivates is

wilhout contradiction, the topics of invention and existence. 3

also able to alter the practices of detection and culture. In other

I )clegation is always an invention in the sense that its point of

words, while the biologist never leaves her lab, she is not locked

departure is not a general project, oriented toward the grand

in. In its own way, the innovative and bifurcating lineage of

l heme of truth, but one that is partial and partisan. When a

laboratories bears witness to a dynamic of apprenticeship for a

"hydraulic door closer" is invented, delegated to automatically

world where "something" is encountered that may create new

close the door behind the visitor to the Center for the History of

problems that challenge the requirements of detection, mea-

Science and Technology in La Villette, it is not acting as a golem

surement, and culture.

or a Turing machine supposedly capable of anything we call

We could say, then, that the Pasteurian biologist, although


dependent on the laboratory, asserts the existence of the microorganism the way the sunflower asserts the existence of the sun,
and Europe, since Columbus, that of America. Of course, the
methods of affirmation are distinct, as are the consequences for
the other term of the relationship: the sun is apparently indifferent to the capture of its photons by plants; the microorganism experiences new adventures butand this is what makes it
an experimental "factish"those adventures, once stabilized,
have, until now, been "explained" in terms of the properties
2

that have been attributed to it as an "autonomous being." As


for "America," its human inhabitants, then gradually its animals, vegetation, water, land, and climate, have been radically
redefined by the history that followed their European "discovery." But beyond those considerable differences, the creation
of significant connections and the affirmation that "there are
worlds," in any case, go hand in hand. On the other hand, the
quantum physicist finds herself enclosed in her lab: there she
is all-powerful because she determines the measurements to be
made, but her measurements seem to prevent her from bearing
witness for the world, something all other creators of connections can do.

"thinking." The door closer is "delegated" to closing the door,


period. It is not replacing a "human in general," but an employee
with a specialized function. This delegation brings into existence an "agent," and this agent is a hybrid being because it testifies both to the interrelations among the nonhuman "actors" it
assembles and the human projects that this assemblage realizes.
And this agent in turn becomes, for humans, a source of new
problems. It can be made to break down if we try to close the
door "ourselves" and, if the hydraulic piston is poorly adjusted,
it will slam the door in the face of a second visitor unthinkingly following the first. Its mode of existence is constantly
being negotiated between two positions: if the device functions
as expected, humans will have to adapt; if not and it "malfunctions," it's "back to the lab." For Latour, the technical invention
must be understood in terms of delegation, of "having X do Y,"
through which both users and devices-agents are invented in
a continual process of redistribution between what belongs to
humans and what is inscribed in the "logic" of the operation
of things. This is "sociological" invention in the strong sense,
which enabled Latour and Strum to contrast human societies
and baboons in terms of the difference between complication
and complexity.'5. Baboon society is complex in that the identity

QUANTUM

6a

IRONY

QUANTUM

IRONY

61

of each baboonwhat it is willingto risk, hope, anticipatemust

who answers. But what is unique to experimental practice is that

be continuously negotiated in real time, with no mark, badge, or

Iliis transformation functions as a critical issue, and as such is

word to stabilize the situation. Our societies, on the other hand,

subject to demanding requirements and challenges. Of course,

are "complicated" in that the majority of our activities satisfy

no requirement and no challenge can eliminate the sleight of

"sociological" prescriptions resulting from successful opera-

hand involved in the transition from the relationship between

tions of delegation that indicate the extent to which one can

I he scientist and her agent to the assignment of a property by the

trust a situation, that allow us to determine "who to talk to," how

scientist to an actor. Pasteur's microorganism is nothing other

to proceed, or with whom or with what we are dealing.


Obviously, the laboratory device entails the success of an

I han the coherent expression of the relationships that Pasteur


was able to create from what he addressed. But it is a witness

operation of delegation of this kind. It operates on behalf of

for the specificity of experimental relationships: these are such

the scientist whose preoccupations it embodies in turn. But

I hat, if successful, they will become part of the proof of the exis-

this operation alone is unable to characterize experimen-

tence of a being endowed with specifiable activities that explain

tal practices; more specifically, it is in their case subject to an

that success. The "experimental factish" is constructed, but the

additional requirement. This requirement corresponds to the

way it is constructed implies that it has to resist the accusation

"realism" invented by the experimental sciences. Not only must

of being no more than a construction, and has to satisfy chal-

the experimental device function as a vector for the creation of

lenges that will test its claim of being endowed with autonomous

a meaningful, information-producing relationship between

existence. That is why experimental achievement is an event.

human and nonhuman, but this relationship must be able to

Delegation has brought into existence a being that can explain

undergo a genuine syntactic transformation so that the informa-

what happens to it in the laboratory and is thus capable of exist-

tion produced can be assigned to one of its poles, the meaning of

ing outside the laboratory.

the relationship being interpreted on the basis of that pole. The

Pierre Duhem pointed out the necessity of resisting the

activity of the delegated agent must have the ability to attest to

temptation to forget that physical theory is an instrument. The

the properties of an actor "independently" of the experimental

rational thermodynamics he defended was distinguished from

project that staged that activity. So Pasteur was able to success-

rational mechanics by the impossibility of moving from its the-

fully argue that living actors, microorganisms, were responsible

oretical object, "reversible displacement," to a reality that would

for the activity of fermentation empirically associated with the

explain the way it is characterized, that would explain "why,"

presence of what were then known as ferments, and he was able

unlike reversible displacement, "it is dissipative." And this

to do so because the making of ferments as his delegated agents

example allowed him to claim that even in rational mechanics, a

included the requirement that the way they would act would

gap must be created between what is subjected to experimenta-

constitute a strong argument in favor of such an attribution of

tion and measurement and the theoretical representation that is

responsibility.

constructed about it. The victory of atoms over thermodynamics

A transformation such as this is not limited to science;

allowed physicists to forget Duhem's claim, but it did not really

we do it all the time, for instance, whenever we attribute the

refute it. The victory of atoms is the victory of devices that can

responsibility for the answer to a question we ask to the one

be used to assign actors and events the responsibility for their

6a

QUANTUM

IRONY

detection, but it remains silent about their "theoretical repre-

QUANTUM

IRONY

63

postmodern thought, is apparently not much favored by phys-

sentation." And when the construction of that representation

icists. Some critics of the Copenhagen interpretation have

occurred, in 1927, it transformed the gap into a prohibition. It

conveyed Schrodinger's protest in a way that expresses their

proclaimed the radical impossibility for theory of justifying

frustration on this point: why can't we, in the quantum case,

experimental practices of delegation liable to bring into exis-

create a syntax that allows us to move from experimental del-

tence what it is the physicist addresses, the impossibility of con-

egation to attribution? Their protest does not contradict Bohr's

structing the quantum being responsible for measurement and

complementarity or deny that the concept to be constructed

of speaking of the laboratory as the place where we "encounter"

relates to practices productive of the information it presents. It

that being and where we learn from it the obligations associated

contests the interruption of the process of construction, without

with a meaningful encounter.

prejudice to the kind of activity we would be led to attribute to

In previous chapters I proposed two kinds of "theoreti-

quantum entities as actors.

cal factish": physicomathematical factishes (for example, the

To take but one example, throughout his life Karl Pop-

Hamiltonian operator) and enigmatic factishes (especially the

per tried to construct their identity in terms of "propensities,"

increase of entropy). To qualify the formalism postidated in

probabilities that owed nothing to our lack of knowledge. For

1927, I now want to introduce a "divinatoiy factish." For it was

Popper, quantum practice entailed the obligation to accept a

indeed a kind of divination that Niels Bohr practiced when he

concept of physical reality in which properties would become

insisted on the fact that the meaning of our measurementsand

"dispositions" (toward this or that), that is, in which the con-

therefore the definition of the "matters of fact" they alloware

cept of "possibility" would obtain the status of an irreducible

relative to "classical" language, "causal" language, a language

category.

that quantum beings do not speak. The wave function together

However, today, an element of novelty has been added.

with its reduction serves as a kind of divinatory apparatus,

Now, the question of the construction of quantum actors could

enabling us to interpret, to confer meaning, to inscribe within

become a practical issue. One representative of these new voices

a practice, "messages" that have arrived from a different real-

is Hans Christian von Baeyer. In Taming the Atom, von Baeyer

ity, that physicists are unable to appropriate, and whose ways

challenged quantum mechanics in a new way. He did not speak

they are unable to penetrate, much less represent. As Bernard

as a representative of a persistent intellectual dissatisfaction but

d'Espagnat has said on many occasions, "reality is veiled," but

on behalf of new devices by means of which the atom had, for

this veil is not a "discovery" that would apply to all humans and

several years already, been presented and interrogated.

would require that they reject what they understand by real-

It is not without interest that the author is the great-grand-

ity. The veil is the counterpart of the formal divinatoiy factish

son of the great chemist Alfred Baeyer, whose name is associated

of quantum mechanics. More specifically, it reflects the irony

with a construction that translates the scandalous success of the

of the message that quantum formalism seems to deliver: it is

"transition" leading from delegation to attribution. Baeyer suc-

about some reality, but it cannot be deciphered as such without

ceeded in representing chemical molecules as commonplace

being, at the same time, cut off from its "source," from "reality."

constructions consisting of balls (atoms) connected to one

Irony, regardless of the praise showered upon it by

another at specific angles by means of rigid rods. Of course,

QUANTUM

6a

IRONY

the rigid rod doesn't tell us anything about a chemical bond any

QUANTUM

IRONY

65

I have chosen a quotation dominated by "subjectivity," for

more than the ball tells us about an atom. But Baeyer's model

II embodies, more than any theoretical expression, the nov-

was the first to inform chemists about the nature of the agents

el ly of contemporary laboratories, where the "microworld" has

they delegate whenever they "make" chemical bodies "react,"

obtained the means to become perceptible in a variety of ways.

and therefore the kind of actor the chemist's practice uses and

III fact, there is no theoretical novelty, in itself, in this becoming

addresses. Could the molecule be conceived as an assemblage in

perceptible. Von Baeyer's mercury atom "responds" to quantum

space? Yes, it could. And it's this "yes" that celebrates the success

expectations, and also corresponds to what was envisioned as a

of delegation leading to attribution, the fact that the assemblage

1 bought experiment sixty years ago. And yet, as von Baeyer's text

satisfies the requirements of the chemist who uses it to imagine

nbows, they expose physicists to a new experience, for which

possible synthesis operations. 5

quantum formalism had not prepared them. It is only indirectly

The great-grandson of the man who "caused" molecules

1 bat quantum mechanics was able to construct the question of

capable of explaining their operations to chemists to "exist" tells

l be atom on the basis of detection experiments involving mul-

us of his excitement when he was confronted with the first device

titudes of atoms (electrons, molecules). But it is the atom itself

that enabled him, not to record the emission or absorption spec-

t bat is, today, in the process of being tamed or domesticated.

trum of the light of a multitude of atoms, but to "see an isolated

In this case, domestication should not be understood to

atom," a "captive" atom, caught in a cavity under vacuum, where

mean something like "good-bye wild freedom, hello sad captiv-

it was held motionless by electrical forces and activated by an

ity." Becoming perceptible means that the atom had obtained

ultraviolet laser. The laser is turned on: "Right in the middle of

the means to impose itself as an actor, and an active one at that,

the trap, a little star appeared. Tentatively at first, amid the flick-

flickering on and off "for real." We should view domestication

ering reflections all around, and then with increasing intensity,

in this case as the beginning of a practice wherein a domus is

the mercury atom poured out its l i g h t . . . So here it was, an atom

created, a suitable habitat, which helps keep the studied being

in captivity. It was then, as I watched spellbound, that I began

"alive," helps "influence" it, multiply the connections with it.

to notice that the atom was blinking. At first I thought that this

The atom of quantum mechanics was a being that had been

was just part of the general flickering of the screen, but it soon

unable to resist its interactions with its milieu from being

became apparent that the mercury atom was definitely turning

absorbed into the theoretical question of measurement. The

off and on, at the rate of several times a second. This was surely

fact of being able "to see it" by means of a subtle manipulation of

the most astonishing thing I had ever seen. Whatever might be

those interactions, through the creation of a "habitat" where the

the cause of this phenomenon, it was a powerful reminder that

atom is able to indicate its individual existence, is less about the

atoms are active, dynamic systems, capable of the most intricate

capture of the atom than the capture of experimental physicists.

internal transformations and convolutions, and not in the least

Von Baeyer is literally captivated by the mercury atom's "impu-

bit like the immutable, eternal kernels of matter the ancients

dent wink."

had imagined them to be. Although I understood this difference


intellectually, it took the impudent winking of a trapped mercury
atom to drive the point home to me in an unforgettable way."7

Today, when untold dollars, hours of work, and subtle negotiation among all the laboratory ingredients have resulted in
an atom that "displays" quantum jumps, that atom has gained

6a

QUANTUM

IRONY

QUANTUM

IRONY

69

itself" have now been resumed "for real." What is at stake is no

I lie atom that flickers in captivity might indeed inspire a shift

longer testing the link between formalism and the prohibition.

in I lie very relationship between the one who questions and the

From the point of view of the industrialist, "behavior in itself"

one who answers. For example, "Where are you?" or "What are

loses its epistemological connotation. It is a question of "mobi-

your energy states?" could give way to the question "What can

lizing" a being we would like to ask to carry out, "on its own," the

we expect of you?" addressed to a potential delegated agent.

operations that interest us. Finally, there is humor in a situation

However, we need to return to theheretofore irresistible

where the "reality in itself" claimed by critics of the Copenha-

ijuestion of the ironic trap constituted by the quantum "divina-

gen interpretation so they might have the right to escape the sti-

lory factish." For the syntactic torsion that transforms a detector

fling confinement of their own questions, the right to describe

into a condition for the actualization of significant information

the laboratory as the place where an encounter with the world

(I he transition from Bohr's first model to the quantum mechan-

is organized, appears in laboratories where technical confine-

ics of 1937) was indeed endorsed by the majority of critics, those

ment is pushed to its extreme limits: the atom and its world are

who initially sought above all to transform the scope and mean-

now confined. The laboratory has become capable of opening

1 ng of that information.

up to the world to the extent that the physicist can delegate to

Here, we can use the present to "relativize" this past. If

her apparatus the role of "world creators" for quantum beings

physicists in 1937 had, like von Baeyer, "seen" a mercuiy atom

that are no longer required to state where they are, or what their

wink at them impudently, I assume they would never have got-

velocity or spin are, but are asked "to act as if we are not there,"

ten interested in a formalism whose primary feature is to deny

that is, to activate themselves within their "own" surroundings.

l he possibility of referring to that wink as an event, as something

Quantum irony has triumphed over its critics. The for-

I hat happens and can be detected. Nonetheless, they had indeed

malism is "complete" and is capable of trapping all those who

"seen" the frequency spectrum of the light emitted by atoms.

undertake to attack it head-on, who try to demonstrate that it is

The radical transformation of theoretical syntax that destroyed

possible to overturn the prohibitions it defines. But "complete"

the possibility of talking about the atom as "detectable" (that

does not imply that it is invulnerable or unchangeable. As Leon

is, as an actor), but only to the extent that it is "detected," may

Rosenfeld repeatedly pointed out, "complete" means endowed

have resolved the difficulties that plagued the Bohr model and

with internal consistency, so that any question it allows finds

resulted in impressive experimental successes. However, it is

answers devoid of ambiguity, and so that all the questions it says

not enough to explain why it imposed itself as if it represented

are impossible actually turn out to be impossible to ask. But

progress without a price, as if the quantum mechanics of 1937

everything depends on what we mean by "asking a question."

spoke the "truth" about the model devised in 1913. In other

Quantum irony is omnipotent in the sense that it goads its crit-

words, quantum formalism, complete and endowed with for-

ics into falling into the trap it has set for them. "What question

midable internal consistency, is also revealing about the value

have you decided to ask?" Anyone who accepts the challenge, by

judgment that helped physicists forget the atom-actor of 1913.

that veiy fact, places herself in the position expected of her. She

The actor of 1913 was a "kinetic" actor. The events that

will, in effect, accept the position of "someone who asks ques-

affected it were characterized in terms of their frequencies,

tions," the position that gives measurement a determining role.

and some of them, notably the spontaneous transitions from

QUANTUM

IRONY

an excited state to a state of lower energy, seemed able to lay


claim to an "acausal" initiative. Like the unstable nuclei found
in radioactivity, these excited states were characterized by a

"lifetime," which is another way of stating, positively, that they


could not be subjected to the requirements of a function, of a
representation that would identify the variables used to predict,
control, or determine the electron's "return" to its fundamen-

The Physicists' Double Standard

tal state.10 Certainly, the Bohr-Sommerfeld atom gave an unexpected role to the dynamics of integrable systems, because the
stationary orbits of electrons were characterized veiy simply
and elegantly by the invariance of the action variable J. But the
Bohr atom didn't obey mechanics. On the other hand, as will be
described later on in greater detail, Schrodinger's wave function
did "obey" a form of mechanicsa modified formbut one that
preserved the essential attributes of Maxwell's Queen of Heaven.
Correlatively, the definition of the observables associated with it
privileged those with a clear mechanical analog (energy levels),
not those that referred to the kinetic actor.

()nce upon a time there was a philosopher. Her name was Nancy
Cartwright and, in 1975, she published, like many others, "her"
analysis of the problem of measurement in quantum mechanics.
This was almost an obligatory step for American philosophers
of science and, like many of them, she decided to proceed on

If, as I speculated, the future of quantum mechanics entails

l he basis of the contrast between the law of evolution to which

the flickering mercuiy atom and the involvement of industrial

the Schrodinger wave function was subjected and the reduc-

labs, it may also entail the question of how the privilege con-

tion of that wave function, which supplies measurement with its

ferred upon the "attributes of the Queen of Heaven" played a role

meaning. Eight years later, that same philosopher published a

in theoretical practices. The laboratory-as-habitat, because it

bookHow the Laws of Physics Liewhose title vividly reflects her

"shows" rather than "detects," gives "that which" is detected the

discovery that she, and all those who accepted the measurement

power to insist. But wouldn't kinetic description, initially dis-

problem as a problem, had been misled.' The last essay in her

qualified as an "incomplete" system of dynamics, then by quan-

book, "How the Measurement Problem Is an Artefact of Math-

tum theory's emphasis on "mechanical" properties also have the

ematics," completes the act of rebellion: the problem that criti-

power to insist? Wasn't a measurable property like "lifetime"

cal philosophers and physicists have been struggling with for

also relegated, along with atomic flickering, to the status of an

generations is a false one.

ectoplasm in the ironic construction of a "matter of fact" that


privileged Hamiltonian facts?" To investigate this further, to
measure the theoretical effects brought about by the seductions
of the Queen of Heaven, will require a bit of technical analysis.

To speak of the "problem" of measurement is to have already


abandoned the Copenhagen interpretation for the one proposed
in 1933 by the mathematical physicist John von Neumann. 2 As
I noted earlier, for Niels Bohr the wave function determined
by the Schrodinger equation lacked all physical meaning

71

72

THE

PHYSICISTS*

DOUBLE

STANDARD

independently of its "reduction," which actualized a choice of


measurement, the wave function then allowing a correspondence to be made between this choice and clearly determined

THE

PHYSICISTS'

DOUBLE

STANDARD

73

very device, if it is considered independently of consciousness, belongs to the same physical reality as the quantum object
Itself.

probabilities of observation. The Schrodinger equation, then,

The problem of quantum measurement, as it is usually pre-

has nothing to do with the description of a given physical real-

Mcnted, begins from the position established by von Neumann

ity, even though it "resembles" the equations of Hamiltonian

l>ut tries to escape his conclusions regarding the decisive role

mechanics, which cause description and reason to coincide.

consciousness would play in reducing the wave function. Tech-

It is part of the physicist's "divinatoiy equipment." Von Neu-

nically, it centers on the contrast between the wave function as

mann's analysis, on the contraiy, contrasts the determinist and

,i superposition of quantum states and a statistical mixture of sys-

reversible "behavior" the Schrodinger equation defines with

tems, each characterized by a single quantum state (and there-

the discontinuous and irreversible event associated with its

to re the possibility of predicting the value of the corresponding

reduction. On the basis of this contrast, he asked: is it possible

observable with probability 1). Measurement on an ensemble of

to expect the equation to provide the secret of its reduction? The

quantum systems results in assigning to each system one, and

answer is no, it's impossible. And it is the consequences of this

only one, value of the measured property. The result appears as a

impossibility that von Neumann's interpretation pushes to the

mixture of distinct, single-state systems. It is as if the measure-

limit. Because any measurement device, including the retina

ment had effected the transition from a "pure case," where all

upon which a photon falls, including Schrodinger's cat, could,

members of the ensemble are described by the same superpo-

in principle at least, be described by a wave function governed

sition of states, to a "mixture" of systems, each with a separate

by the Schrodinger equation, none can be held responsible for

identity, that is, each described by a single-state wave function

its reduction. Then, what is it that makes something a mea-

corresponding to one, and only one, value of the observable.

surement device? Von Neumann saw only one place where his

(:an we give a theoretical interpretation to this transition, which

regression operation could be stopped, where what appeared to

cannot be understood in terms of the Schrodinger evolution?

be responsible for the reduction would not in turn prove to be

"Quantum mechanics requires a superposition: the philo-

itself subject to reduction: this would be the consciousness of

sophical problem is not to replace it by a mixture, but rather to

the observer.

explain why we mistakenly believe that a mixture is called for."4

Bohr had spoken of a logic embodied in a concrete device,

This is the way Nancy Cartwright presented the problem in 1975.

one whose characterization includes its purpose and, therefore,

She indicated how she wanted to see it resolved: the "transition"

the way it will be used. His favorite example was that of a stick

would be explained away as an illusion. The Schrodinger evolu-

held in the hand. Depending on whether it is held tightly or

tion, whose subject is superposition as such, would govern the

loosely in the hand, the stick satisfies two logically incompatible

situation. To construct this solution, it is necessary to accept,

definitions concerning the information it will provide by tactile

like von Neumann, that the Schrodinger equation has unlimited

exploration. Thus, we are forced to choose between two mean-

validity and, in particular, obligates us to define the measure-

ing-producing uses. As for von Neumann, he associates the

ment instrument in terms of superposition. But, contraiy to

finality of the device with pure conscious choice, arguing that

von Neumann's position, the analysis would not be reduced to

75

THE

PHYSICISTS*

DOUBLE

STANDARD

the axiomatic claim that the atom and the measurement device,

THE

PHYSICISTS'

DOUBLE

STANDARD

73

1 ,inc or a mixture. The "reduction" of the wave function would

whenever abstracted from its aim, are the same thingjust more

then be nothing other than the result of the fact that the evolu-

complicated. The crucial point would be to explain why measure-

tion of the "measuring system" toward equilibrium allows an

ment results seem to imply the destruction of superposition, the

i|proximation that eliminates all information capable of indi-

reduction of the wave function to a mixture of single-state wave

.iting the difference between a superposition governed by the

functions. "Why do we maintain the fiction that we can assign a

logic of "and" and a statistical ensemble governed by the logic

fixed value to an observable, even a well-constructed one?

o! "or." We are mistaken if we think that, once measurement

From this point on, measurement becomes a physical prob-

has taken place, we can assign one, and only one, quantum state

lem, an interaction whose specificity needs to be defined. As

to each quantum systemthat is, whenever we think that the

Leon Rosenfeld wrote, "we should represent measurement as an

measurement has transformed the pure case into a mixture, or

interaction between the observed atomic system and a record-

whenever we think that our "measurement" allows us to associ-

ing device, an interaction that ultimately leaves some perma-

ate one, and only one, value to what we measure. But, concluded

nent mark unequivocally associated with a definite quantity that

(iartwright in 1975, this error is of no consequence. It will never

characterizes the state of the quantum system." Speaking about

lead us to make a prediction that would contradict a possible

a permanent mark points to a way of defining the specificity of

measurement because our measurements are, by definition,

the measuring device. It should be such that the measurement

macroscopic, and relative to the state of equilibrium attained by

interaction can be described in terms of an "evolution toward

I be measuring device.

equilibrium." Therefore, it is to statistical mechanics, a science

What happened so that, a few years later, Nancy Cartwright

that articulates equilibrium in the thermodynamic sense with

would reject what is in effect a conciliating solution? Why did

dynamic or quantum description, that we might turn for the

nbe suddenly no longer feel "obligated" to give the wave function

secret of the apparent reduction.

I be power to produce the theoretical understanding of mea-

Like Cartwright, Rosenfeld referred to the results proposed

surement? In the interim she had taken a highly unusual step

in 1965 by three Italian physicists, Adriana Daneri, Angelo

lor a philosopher. She examined not only the literature in which

Loinger, and Giovanni Maria Prosperi. They had shown that

problems of interpretation were discussed, but also the texts

when considering measurement, now qualified as "macro-

that transmit the procedures accepted by the community and

scopic," as an evolution toward the equilibrium of the "measur-

from which physicists learn their craft. She came to an aston-

ing system," whether the point of departure for this evolution

ishing conclusion. When the specialists of quantum mechan-

was a "pure case" or a "mixture" made no discernible differ-

ics were no longer concerned with the fundamentals but were

ence. From the moment the relative proportions of the differ-

building the "real" foundations of quantum mechanics, that is,

ent systemseach characterized by one, and only one, quantum

its ability to occupy its corresponding experimental terrain, they

stateconstituting the mixture correspond to the respective

no longer felt obligated by the mathematical properties of quan-

probabilities of the possible measurements that can be calcu-

tum superposition. More specifically, they used those properties

lated from the wave function for the pure case, the result of an

for part of their calculations, but, at a given moment, introduced

ensemble of measurements will be the same whether for a pure

a judicious approximation that destroyed the superposition.

77

THE

PHYSICISTS*

DOUBLE

STANDARD

THE

PHYSICISTS'

DOUBLE

STANDARD

73

Worse, Cartwright discovered, quantum theory did not establish

The "lie" associated with physical laws indicated by the title

the point in the calculations at which the approximation could be

iif (Millwright's book indicates that she has discovered that, even

introduced. The physicists were guided by experimental data.

in classical dynamics, "laws" do not explain phenomenological

In fact, Cartwright found that the "reduction of the wave

irgularities. In all concrete cases, the law requires additional

function" is not even specifically associated with measurement.

delails that provide it with its relevance for experimental facts,

Physicists "need" the reduction of the wave function not only

IIIII the crucial nature of those details, which the law needs but

when they make a measurement but every time they "prepare" a

does not justify, is "forgotten" whenever we claim that the law

system, whenever, for example, they produce a beam of electrons

"explains" the fact.

in a linear accelerator that are supposed to be characterized by a


clearly determined quantity of motion. "This kind of situation
occurs all the time. In our laboratories we prepare thousands
of different states by hundreds of different methods eveiy day.
In each case a wave packet is reduced. Measurements, then, are
not the only place to look for a failure of the Schrodinger equation. Any successful preparation will do."6 Moreover, this failure
occurs whenever it is a question of accounting for a process that
presents the problem of its preparation rather than its detection. For instance, when physicists produce laser radiation,
they make use of equations suggesting that transitions "actually" occur, similar to events that are indifferent to whether
or not they are being observed. And if those equations, which
are kinetic in nature, are to characterize events, the interference terms resulting from the quantum superposition have to
disappear, as in the reduction of the wave function. Of course,
those terms are necessaiy for correctly presenting the problem,
Cartwright notes, but their disappearance, through the intervention
of an appropriate approximation, is just as necessary if the problem is to have a solution. In other words, for the vast majority
of "experimental confirmations," which add their prestige and
authority to quantum laws, there is no theoretical deduction at
all of the value of a measurable property that would confirm the
measurement. Rather, there is negotiation between the way the
"laws" present the problem and the type of solution required by
the experimental data, between theoiy and "phenomenology."

Cartwright recognized the affinity of her position with that


of Pierre Duhem, for whom theoretical structures were, above
.ill, instruments for ordering and classifying experimental
accounts, but she doesn't follow him in his fight against theoretical entities themselves. This accurately reflects the transformation of phenomenological and experimental practice within
l he century. Atoms, electrons, and protons "really" exist in the
sense that they are not simply presented by theoiy but are agents
delegated by practice. When Rutherford subjected a gold leaf to
l he "radiation" from a radioactive source, there was no possibility of understanding what he obtained other than by introducing entities capable of causing change and undergoing change,
I hat is, no possibility of describing what he did other than as
a showering of his target with discrete entities that, under the
circumstances, behaved like tiny projectiles. For Cartwright, an
entity used in an experimental device to provoke or create the
new phenomenon being studied obligates the experimenters,
by that very fact, to endow it with existence.7 "Laws," as Duhem
claimed, "save" phenomena. But, Cartwright adds, most likely
because, unlike Duhem, her physics is populated with detectors, the entities such laws introduce must be recognized, independently of those laws, as integral parts of the phenomenon.
And because of those detectors, physicists and electrons coexist
in the strong sense, the way the plant coexists with the sun, or
Europe, since Columbus, with America.
Cartwright recognized the autonomous existence of what I

79

THE

PHYSICISTS*

DOUBLE

STANDARD

THE

PHYSICISTS'

DOUBLE

STANDARD

73

call "experimental factishes," beings whose existence experi-

like all the laws of physics, they invite lies. Not only do they

menters accept because they are able to make them do things in

i irate the temptation to lie, as we have seen, they appear to

the laboratory. The physicist who manages to delegate "puts into

impose it upon those who refer to them. For, the physicist "at

practice" an "agent" and, therefore, also brings into existence

work" who attempts to describe laser radiation or design a del-

an actor, if the agent has the power to present the physicist with

egation device must use the Schrodinger equation "as i f " it had

problems "of its own." On the other hand, Cartwright feels free

descriptive value, but is also forced to cheat, to have recourse to

to consider as "liars" what I have called "theoretical factishes"

.ipproximations that enable her to clandestinely deny what the

(physicomathematical, enigmatic, and divinatory). These inter-

e<| nation asserts openly.

vene whenever theories "judge" experimental practices and

Cartwright introduces a problem that must be understood in

refer them to a "reality" that assigns them one role, that of access

terms of obligations. If physicists at work are not obligated by

to its ownnow discovered(theoretical) truth. Of course, there

quantum laws, if they feel free to introduce the needed approxi-

are so many causes to be defended and value judgments encum-

mations, why should those laws claim an authority that would

bering and shaping this assignment that it seems possible to

create obligations for others, for philosophers and for those

associate it with a merely social mode of existence. But with her

who are mesmerized by the question of knowing if and when

notion of the "lie," Cartwright may be overlooking the fact that

Schrodinger's cat "really dies"? What we hear in the accusa-

those "factishes" are themselves capable of posing problems of

tion oflying is not the repetition of Schrodinger's "cry" when he

their own to those who bring them into existence. More specifi-

insists that the cat really dies. His cry dramatizes the disastrous

cally, her position makes her vulnerable to the temptation to dis-

shortcomings of Cartwright's 1975 solution, for that solution

qualify such problems for being false problems. It might be better

amounts to saying: what "happens" to the cat, a macroscopic

to say that it is a question of entirely different types of factishes

measurement instrument, does not matter; what matters is that

and of their respective modes of evaluation. Experimental del-

what we call the cat's "death" is a well-constructed approxima-

egation, which seems to be part of a concoction made in a labo-

I ion; what matters is that the terms that translate the fact that

ratory kitchen, may ultimately interest everyone through the

I he "dead" cat is in superposition with a living cat make no mea-

technical devices that continue its practice, while the theoretical

surable difference statistically. The accusation of lying, rather,

factish, which is assumed to interest everyone, which is offered

follows from the realization that the whole question of the cat

as the focus of the public's (and the philosopher's) fascination,

dissolves once we turn to what physicists do at work, construct -

gets its power solely from the passion that singularizes a physics

i ng tools and accepting obligations entailed by questions far less

in which Duhem appears as the vanquished.

"strident" than the question of what happens to the cat.

I want to return to the case of quantum laws, where Cart-

It is by presenting the contrast between, on the one hand,

wright can indeed speak of lies because she has been forced

the "quantum event" physicists at work continue to have need

to take sides by the way these laws were presented. For Cart-

ofespecially that "primordial" event, the decay of an excited

wright, the "philosophical" problem of measurement is a false

atom, by which the atom forces us to recognize its nonclassi-

problem, an "artifact" purely and simply, created entirely from

cal electronic structureand, on the other hand, the "simula-

mathematical formalism. Those laws are not limited to "lying";

crum" to which that event corresponds in post-1927 quantum

81

THE

PHYSICISTS*

DOUBLE

THE

STANDARD

PHYSICISTS'

DOUBLE

STANDARD

73

mechanics, that Cartwright answers Schrodinger's "cry." There

iilence of the one who "knows very well, but all the same." The

is no need to cry about the cat; rather, let us ciy about the absur-

I rap no longer functions with its two-stage irony, creating the

dity of this simulacrum, about the absurdity of the theoretical

temptation to use the Schrodinger function to "describe," then

story of a decaying atom.

disallowing the description. Cartwright has entered the physi-

"On this stoiy nothing happens. In atomic decay the atom

cists' "kitchen," she has read what philosophers don't normally

begins in its excited state and the field has no photons in it.

read, the "technical details." She refers to their practices rather

Over time the composite atom-plus-field evolves continuously

I lian to the "masterpieces" they expose in public.

under the Schrodinger equation into a superposition. In one

In order to deny any physical meaning to the Schrodinger

component of the superposition, the atom is still in the excited

equation and relate it to an ascetic encounter with the experi-

state and there are no photons present; in the other, the atom is

mental data that are the sole source of legitimacy in physics,

de-excited and the field contains one photon of the appropri-

I hose data must correspond preciselyand here the cases are

ate frequency. The atom is neither in its outer orbit nor in its

assimilable to "masterpieces"to the privileged observables

inner orbit, and the photon is neither there in the field travel-

that are the quantum analogs of Hamiltonian variables. But the

ing away from the atom with the speed of light, nor absent. Over

fecundity of quantum mechanics makes reference to all kinds

time the probability to 'be found' in the state with an excited

of situations that offer no such analogy. And it is for such situ-

atom and no photons decays exponentially. In the limit, as

ations that, notwithstanding the Copenhagen interpretation,

t > o, the probability goes to zero. But only as t >

physicists have to behave "as i f " the Schrodinger equation did

On the

new-quantum-theory stoiy, never, at any finite time, does an


atom emit radiation."

Of course, this cry could be silenced by a physicist who


responded to Cartwright that time as presented by the
Schrodinger equation has nothing to do with any physical temporality associated with the phenomenon. It's relative, like
everything else, to the time of measurement. This answer would
be that of an authentic defender of the Copenhagen interpretation, for whom the problem of the "reduction" of the wave
function is a false problem because the wave function doesn't
correspond to any story, is nothing without its reduction. Quite
simply, Cartwright doesn't have the right to use the Schrodinger
equation to "narrate" what happens between the initial moment
(excited atom) and infinity (photon emitted). But this time, and
this is what is interesting about Nancy Cartwright's position,
the discussion doesn't end here, with the silence of its protagoniststhe "ascetic" silence of Bohr's disciple, and the frustrated

indeed describe an objective situation. They preserve the syntax


of the equation, which prevents them from bringing into existence the atom, the field, or the electron independently of measurement, but use it to introduce agents that they treat as if they
existed independently of measurement. It is in the course of this
delicate operation that the "kitchen concoction" of physicalmathematical approximation intervenes. It will enable the situation they interrogate to preserve, where it must, the relevant
quantum features, and, again where it must, to conveniently
ignore others as insignificant. For example: approximation will
result in assigning a lifetime or a cross section to a particle, so a
kinetic description can be provided, or approximation will allow
a clearly defined role to be attributed to an agent. Whenever they
need to create an experimental situation where a quantum being
is "delegated" to causing events, physicists no longer feel obligated by the asceticism they propose, to the fascination of philosophers. Physicists are playing both sides of the coin.

83 T H E

PHYSICISTS*

DOUBLE

STANDARD

That's why Nancy Cartwright dares shift her position from


that of a philosopher, who comments and tries to understand,
to that of a "physicist," in the sense that one must be a physi-

THE

PHYSICISTS'

DOUBLE

STANDARD

73

momentum, or some other conserved quantity, that something


happens in quantum mechanics." 9
It is worth remembering that Bohr asserted the "acausal"

cist, in a technical sense, to be part of the controversy with other

nature of quantum physics, but he gave the term "cause" the

physicists. For her, quantum language is not only "deceitful" in

Name meaning it has in dynamics, one in which change observes

the sense that, in general, all theories lie, it is "distorted" in the

1 he conservation of cause in effect. When Cartwright speaks of

sense that it is prone to present "bad" observables as signifi-

1 he "causal role" that allows us to refer to a property as real, she

cant. We make use of a screen to claim "the electron is here," and

emphasizes the concept of a "role." She asks that the physicist

the reduced equation seems ready-made to provide us with the

who brings experimental devices into existence construct a

probability that it is actually there. By recording a photon "emit-

coherent theoretical language capable of "telling stories" about

ted" by an excited atom, we can claim that the atom is now in its

what happens. In other words, here, the meaning of a cause is

stationaiy state, and in the Schrodinger equation the probability

found in the practice of delegation, in the staging that simultane-

amplitudes corresponding to the different states are effectively

ously brings into existence an agent, the stoiy of its production,

in superposition. But we "distort" measurement when we use

and that of its action. From this perspective, dynamic "cause"

it this way, making it bear witness to something about which it

becomes a unique case, which structures "stories" whose com-

is, in fact, silent. Measurement never bears witness, Cartwright

mon feature is the role they give to the = sign. But these stories,

notes, to a "state" defined in terms of energy, position, momen-

like all others, belong to the process of delegation. Galileo asked

tum, or any other observable to which Schrodinger probability

I he rolling ball to bear witness to the fact that, from the view-

amplitudes for a state would correspond. It witnesses transi-

point of what its motion makes it capable of, the only thing that

tions, and associates with transitions properties, such as life-

counts is the change of height that occurs during that motion.

times, or cross sections (in the case of scattering experiments),

In Hamiltonian physics, the extraordinarily simplified nature of

that do not belong to the same physical syntax as the observables

the "stoiy of what happens to the ball" is used to make its mono-

to which the definition of quantum states corresponds.


Why, Cartwright asks, has this tension gone unnoticed by

tonic nature explicit: there exists a point of view that allows us to


state that, no matter where the ball goes, "nothing" happens to it.

critical philosophers? "They want to find some way to ensure that

The tension Cartwright points out between the "causal role"

a quantum system will possess values for all the classic dynamic

quantum experimental devices confer on energy exchanges,

quantitiesposition, momentum, energy, and the like. But this

such as the spontaneous transition of an excited atom to its

motivation is ill founded. If we want to know what properties are

Fundamental state (which "causes" a photon to be emitted), on

real in a theoiy, we need to look at what properties play a causal

the one hand, and the theoretical syntax of quantum mechan-

role in the stories the theory tells about the world. This is a point

ics, focused on variables analogous to those of dynamics, on the

I have been urging throughout this book, and it makes a crucial

other, is even greater than she describes. Take the mathematical

difference here. Judged by this criterion, the classical dynamic

simulacrum of atomic decay as Cartwright characterizes it: what

quantities fare badly. The static value of classic dynamic vari-

the Schrodinger equation "describes" is nothing other than the

ables has no effect; it is only when systems exchange energy or

time evolution of the superposition of the excited-atom-plus-

85

THE

PHYSICISTS*

DOUBLE

STANDARD

THE

PHYSICISTS'

DOUBLE

STANDARD

73

field-without-photon and de-excited-atom-plus-field-with-

1 ill> existence. That's why I have chosen to discuss the approach

photon, a photon whose frequency can be measured. But such a

lnkcn by quantum mechanics in reverseby not supplying the

simulacrum is already itself a compromise solution. If the super-

necessary technical elements" and moving directly to their

posed probability amplitudes evolve overtime, it is only because

consequences for each and every one of us." I wanted to begin

the initial representation involves an atom described as if it

with the "grand problems" and end up by asking, "What game

were "perturbed" by an electromagnetic held, that is, with ref-

nre physicists playing?" The moment has come to describe the

erence to an atom that would be "truly isolated." Such an atom

.1 range (official) rules of that game.

corresponds to a Hamiltonian, Ho, in the Schrodinger equation,


and its energy levels will be characterized in terms of constant
probabilities over time. In other words, it will never decay. The
fact that the equation for the atom perturbed b j the field allows us
to define the evolution of an "emission" probabilitythe probability that a photon will be measuredis therefore directly
dependent on the way the problem has been presented. It is
because the interaction between atom and field is defined as a
disturbance of the atom by the field, and not as a unique system defined by a Hamiltonian that would include the held, that
the probabilities evolve over time. And this choice of definition, even if in general it is the only one practical, is nonetheless
"only practical" from the point of view of quantum mechanics.
Indeed, from this point of view, "it should be possible" to construct the Hamiltonian of the "atom-plus-field" system in such
a way that, like the isolated atom, it is represented by a superposition of stationary states each of which evolves individually in
what is called Hilbert space.10
Until now, I have limited myself to claiming that quantum
mechanics had extended Hamiltonian dynamics by transforming it, but I didn't specify what that transformation consisted of.
That I have now introduced the quantum Hamiltonian and Hilbert space indicates that I need to say (just a little) more about
them. In a held such as physics, the telltale sign that questions
are gradually gaining in relevance is always the same: it is found
in the transition from "ideas" to "technical details," that is, the
singular mode of existence of the beings the physicist brings

THE

SILENT

DESCENDANT

8?

wave function, therefore, could not even be raised within the


context of Bohr's thought experiments. 1 That is why these exper-

iments helped give the quantum "divinatory factish" a scope that


was limited only by the "consciousness" of the observer. I now

The Silent Descendant


of the Queen of Heaven

want to show, referring to von Neumann's axiomatic formulaI Ion, which provided quantum formalism with its most purified
form, what this unlimited scope means. In writing the simple
(.reek letter tp, the physicist makes claims compared to which
1 hose of Laplace's demon, who observes the universe as a system
nl' interacting masses, are trivial. 2
At the center of this axiomatic formulation, which extends
iind transforms the dynamic Queen of Heaven, we find a fundamental couple, the wave function we have already encountered,

Let there be a quantum system represented by a wave function

;ind the quantum operator, which "acts" on that function.

ip. This statement, which is extremely commonplace, can refer

The operator is a new being that was invented by quantum

to the universe as a whole in certain areas of theoretical cosmol-

mechanics but is not limited to it. So, for today's physicist, every

ogy as well as to "Schrodinger's-cat-unstable-atom-poison-

lime an evolution equation unequivocally associates the varia-

triggering-device." It's as if the statement were free, presenting

1 ion over time of a function at a given moment with the value of

no problem of relevance, as if we could require any portion

1 he function at that moment, the connection determined by the

of reality to respond to a quantum description. It is impor-

equation can be translated into an "operator/function" formal-

tant to recognize that, in his own way, it was Niels Bohr who

ism. We could then say that the evolution over time of the func-

led quantum physics down this rather strange rhetorical road.

l i o n / i s given by an operatorlet's call it Gacting on/, which

Von Neumann's axiomatic interpretation, by which any mea-

is then written: df/dt = Gf. In general, as its name indicates, an

suring instrument in fact, anything (except the comprehend-

operator "operates" on a function, that is, transforms it into a

ing mind)must be able to be represented by a wave function,

different function.

followed the path of Bohr's thought experiments. Bohr tried

The creation of the concept of an operator in quantum

to show that the identity of a measuring instrument inevitably

mechanics translates, in mathematical form, the difference

implied a choice, a way of using that instrument, and that the

between the "representation" of a system by a functionwhich

complementary measurements satisfied logically contradictory

is the case for classical mechanicsand the ability to character-

methods of use. To do so, the physicist had to behave "as i f " what

ize that system with observable values using a wave function.

was to be the quantum object, represented by a wave function,

In quantum mechanics, there is an operator for each observ-

and what was to be the instrument responsible for the reduction

able, but this correspondence has a precise physical meaning

of the wave function remained perfectly indeterminate until the

only if the wave function on which the operator acts is itself

moment of choice. The question of the limits of relevance of the

expressed in terms of what are known as "eigenfunctions" (or

86

THE

SILENT

DESCENDANT

THE

S RL E N T D E S C E N D A N T

97

"eigen states") of that operator. The particularity of these eigen-

I lie set of these numerical coefficients corresponds to the well-

functions is that the operator of which they are eigenfunctions

I uown "probability amplitudes" that must be "reduced" so the

does not transform them into other functions. The action of the

respective probabilities of the different possible results of a

operator on one of its eigenfunctions yields that same function

measurement can be calculated. This reduction is obtained by

simply multiplied by a number. That number is an eigenvalue of

squaring the numerical coefficients associated with the eigen-

the operator. And it is these eigenvalues, finally, that supply the

limctions of the superposition. As a result of this operation, the

measurable values of the observable with which the operator is

physicist is able to associate each eigenvalue of the operator cor-

associated.
The quantum wave function or, more precisely, the superposition of eigenfunctions by which it can be represented, is

1 rsponding to her measurement with a probability, that is, she


in able to determine the respective probability of the different
values this measurement will produce.

defined as an element of "Hilbert space." 3 For von Neumann,

For each operator chosen, the wave function is represented

this Hilbert space, which quantum functions "inhabit," has

hy a superposition of eigenfunctions (or states), each of which

become the true subject of quantum theoiy. We could say that

is associated with two properties of interest to physics: the

quantum operators transform an element of Hilbert space into

probability amplitude and the eigenvalue. Choice of measure-

another element of Hilbert space. The way in which they are able to

ment/choice of operator/choice of representation in terms of a

transform it comprises the rules of the quantum game, the syn-

iiiperposition of eigenfunctions in Hilbert spacethey are three

tax that governs its statements, that is, determines those state-

aspects of one and the same choice. Independent of this choice,

ments to which meaning can be assigned.

l he wave function is silent.

What's the relationship between this syntax and the ques-

And yet, the wave function is not entirely silent, for we

tion of "choice of measurement" that served as Bohr's leitmotiv?

know it is governed by the Schrodinger equation, which deter-

We accept, axiomatically, that a system's "quantum state" is rep-

mines its evolution over time. It is here that we encounter an

resented by a i[> function in Hilbert space. As such, this function

old acquaintance, the Hamiltonian, which has now become

is physically silent. It has a mathematical existence, but it says

the "Hamiltonian operator." And it is with the definition of the

nothing about the possible results of measurements that will

I lamiltonian operator that an axiomatic formulation apparently

actualize the physical existence of the system. In order for the

devoid of "physical meaning" (the four previous paragraphs

function in question to be able to "speak," it must be defined as a

have only proposed formal definitions) assumes a very precise

superposition of eigenfunctions of the operator that corresponds

meaning. Even if, from the ascetic viewpoint of the Copenhagen

to the proposed measurement. The "choice" of measurement

school, Hilbert space, which shelters the Hamiltonian operator,

instrument, therefore, corresponds to the "choice" of repre-

may indeed have only mathematical significance, nonetheless it

senting x|) in terms of this (infinite) set of eigenfunctions rather

hears a message that has the power to lure physicists away from

than in terms of another family of eigenfunctions. 4 More spe-

I hat point of view. How could they resist conferring upon Hilbert

cifically, the "state" in question will be represented by a super-

space the status of an autonomous "factish"? For it is the direct

position of eigenfunctions of the operator, each superposed

descendant of Hamiltonian dynamics and, moreover, belongs

eigenfunction being "weighted" by a numerical coefficient.

to the line of the royal family itself. Hilbert space is not limited

96

THE

SILENT

THE

DESCENDANT

SRLE N T D E S C E N D A N T

97

to claiming the power of equivalence, that is, of governing how

reality itself, to supply (Hamiltonian) reasons for the way reality

the problem of quantum evolution is constructed. The Hilber-

behaves. But physics is not a moral tale, where liars are always

tian definition of quantum beings, eigenfunctions and opera-

punished. The creation of problems is what matters, whatever

tors, implies the ability to characterize any "quantum system"

I lie starting point.

in terms suitable only to integrable systems as understood by

Let's take a closer look at the Hamiltonian operator, which

Poincare, that is, an ensemble of autonomous entities evolving

makes Hilbert space a part of the prestigious lineage of the

over time independently of one another.

"cyclical" form of dynamic representation where, as we saw in

In other words, von Neumann's axiomatic formulation ele-

Book II, power, intelligibility, and beauty converge. In a sense, it

vates the ambiguous irony of quantum formalism to a nearly

is just one operator among many. Its eigenvalues correspond to

perverse level. For Hilbert space, where "quantum states" can

the measurement of energy, for example, the energy of the dif-

be represented, irresistibly calls forth a presentation of the

ferent stationary states of the Bohr atom. However, in dynamics,

situation where the wave function appears as "representing" the

energy is not a property among others. Its dynamic (not ther-

quantum system. A "Hilbertian realism," which assimilates Hil-

modynamic) conservation reflects the relationship by which,

bert space to a "physical-mathematical truth" that transcends

ever since Galileo and Lagrange, instantaneous states and causal

the phenomenal world the way Einstein's four-dimensional

equality mutually define each other. The Hamiltonian dynamic

space-time transcends our world, where space and time are

assigns this relationship of mutual definition the authority of

separate, could, therefore, almost be said to be dictated to the

a syntactic rule expressing the veiy identity of the system. The

physicist (and the philosopher). We can now "forget" the ascetic

Hamiltonian is at the same time that which eveiy canonical

stipulation passed down from Copenhagen: abandon all repre-

transformation of the representation of the system must main-

sentation, Hilbert space is purely mathematical; physically, it

tain as invariant and the key property of the system's evolution

is nothing without the reduction of the wave function. We can

equations. Because it is kept invariant during the evolution, this

forget it because physicists continue to forget it the way one for-

evolution can appear as a simple transformation of representa-

gets the "voice of one's conscience""Yes, I know, but all the

tion: the system "is," the point of view from which it is repre-

same." Here, my purpose is not to relay this voice and admon-

sented "changes." Because the Schrodinger equation links the

ish physicists. Rather, I wish to follow the way in which prob-

evolution over time of the wave function with the action of the

lems that currently fascinate them have been constructed. We

Hamiltonian operator on that wave function, the Hamiltonian

must acknowledge that physicists have indeed been "captivated"

operator inherits this critical role, enabling the identity of the

by Hilbert space and that the way they have used itin spite of

system and evolution over time to communicate systemati-

the admonitions from Copenhagento state their problems

cally.5 With respect to the question of determining how a quan-

(one example, as we have seen, is the "problem of measure-

tum system can evolvemore precisely, how the probabilities of

ment") have turned it into a physical-mathematical "factish,"

the different results of measurements to which it is subject can

a physical-mathematical being that is acknowledged to possess

evolveit is the superposition of eigenfunctions of the Hamilto-

an autonomous existence. And we can indeed say that this fac-

nian operator that will provide the answer.

tish is a "liar." For it pretends to speak the language of quantum

However, as I have stated, the power of the Hamiltonian

96

THE

SILENT

DESCENDANT

THE

S RL E N T

DESCENDANT

97

operator is greater than that of the classical Hamiltonian. The

lie represented as a superposition of eigenstates (or functions)

primary goal of classical dynamics is to construct the equations

of the Hamiltonian operator. And whenever this is the case, the

of motion for a system. In some cases, those equations will be

Schrodinger equation will describe an essentially static situ-

able to be integrated (resolved) and the spatiotemporal defini-

ation. The evolution of the wave function in Hilbert space can

tion of trajectoiy will then be given explicitly. It is in these cases

lie compared to the simple, monotonic rotation of the different

alone that the dynamic system can be given a cyclical represen-

eigenfunctions of the Hamiltonian it superposes. Consequently,

tation, can be represented by equations of motion in which each

l he probabilities corresponding to the eigenvalues of each of

of the system's degrees of freedom evolve independently. In this

l hose eigenfunctions are, by definition, constant overtime.

case, it is no longer energy alone but each of the momenta (or

In other words, the discrete nature of the energy spectrum

quantity of motion) characterizing those degrees of freedom

I hat characterizes quantum beings has been given an extraordi-

that become an invariant of motion. But in quantum mechanics

narily powerful meaning by quantum mechanics. If a system has

the situation is different because motion invariance was the very

a discrete energy spectrumand this property has become uni-

instrument used in the primordial capture of quantum beings

versal ever since Planck's constant acquired the status of a uni-

by the Hamiltonian formalism (see chapter 5). Each stationary

versal constantit must be able to be represented in terms of the

state of the Bohr atom was already seen to be characterized by

eigenfunctions of the Hamiltonian operator/' Therefore, it must

an invariant of "electronic motion," the value of the different

be able to be represented in a way that causes interactions to

invariants in this case forming a discrete spectrum rather than

disappear and allows the system to exhibit periodic behavior in

being, as in classical dynamics, a function of the initial state of

Hilbert space. The benefit of this representation is the answer it

the system (where each degree of freedom of the system can

gives to the question of evolution over time. For there is one and

assume any value). And it is this mode of characterization that

only one answer: "Nothing is happening."

will be extended and systematized by the definition of the wave

I want to introduce one last technical term, which will

function in Hilbert space. In quantum mechanics, to state "let

become crucial at the end of Book V, namely, the concept of spec-

there be a quantum system represented by a wave function ip" is

tral representation. Just as the Hamiltonian of a classical integra-

to assert that this system can be represented as a superposition

t e system can be written in terms of variables corresponding to

of stationary "states," each evolving on its own, independently of

its cyclical representation, the quantum Hamiltonian operator

all the others, over time. And this is so because whenever time

can be defined in terms of its eigenfunctions and eigenvalues.

evolution is involved, the system will be represented in terms of

This is the "spectral representation" of the operator. To be able

the eigenfunctions of the Hamiltonian operator.

to explicitly write the spectral representation of the Hamilto-

This, then, is the significance of the Schrodinger equation,

nian operator of a system is to be able to state a problem using

which introduces us to the action of the Hamiltonian operator

the very terms of what will be the solution of that problem: the

on a wave function. And its scope is shown by its unbridled use

problem itself affirms the convergence between power, intel-

by physicists when they claim that Schrodinger's cator any-

ligibility, and beauty. But the spectral representation of the

thing at allis subject to its strictures. For this equation (and the

Hamiltonian operator eliminates any possibility of assigning an

wave function itself) to have meaning, the wave function must

intrinsic meaning to the concepts physicists need. There is no

96

THE

SILENT

DESCENDANT

possibility of defining the lifetime of an excited atom in those


rare cases when the spectral representation of a system can be
explicitly defined. "Nothing happens," except, as we saw earlier
(chapter 5, note 10), the ghostly dissolution of an initially concentrated wave packet, the atom and its discrete energy levels
having been eliminated by the spectral representation.
It is fortunate, therefore, that spectral representation is
not generally accessible in cases where the physicist knows
that "something is happening." In fact, every time physicists'
practice leads them to study a quantum situation where things
"happen" (the emission or absorption of a photon, collision,
acceleration in a freld, laser emission), they do not try to provide the situation with a spectral representation where precisely
nothing would happen. In such cases, physicists turn to the
technique of perturbation. They start from the spectral representation of the "unperturbed" system and "add," in the form of
perturbations, the interactions that will be responsible for what
"happens." These interactions couple the eigenstates of the
unperturbed Hamiltonian, states that are then no longer stationary. Yet, quantum mechanics is categorical. It requires that
every "perturbation," every coupling between eigenstates of the
Hamiltonian operator, be defined as eliminable, resulting in a
new Hamiltonian and new independent eigenstates. The probabilities that vary over time and allow us to construct the physical meaning of properties such as the lifetime of an excited state
have, therefore, a meaning that is purely dependent on the fact
that physicists choose not to take the requirements of the formalism to their logical conclusion.
Consequently, quantum mechanics imposes the use of a
mathematical structure that will never provide intrinsic meaning to observable properties in terms of what Cartwright refers
to as the "causal role" of energy exchanges. This role is inappropriate for the inhabitants of Hilbert space. Their presentation,
whenever that role is being performed, whenever it's a question,

THE

SRLE N T

DESCENDANT

97

1'H example, of constructing the kinetic equations that describe


laser emission, requires "leaving" that space, a theoretically
unjustified operation that, as we have already seen, is realized
lit rough the relevant approximations. As Cartwright noted,
physicists are forced to "cheat," to introduce approximations
I ha I allow them to alter the mathematical structure of the problem. Those approximations clandestinely provide the coupling
perturbation with a "truly causal" role, one that cannot be written off by the transformation that would eliminate the coupling
by incorporating the perturbation in the Hamiltonian.
More generally, there is no actualization of the quantum
object defined in Hilbert space without an exogenous" intervention, an interaction that is not described in Hilbertian
terms. This results in the "theoretical violence" that von Baeyer
complained aboutthe mutilation rather than the negation of
practice. For it is this method of definition that relates everything that "happens" to the intervention of measurement, to the
recording of a datum by an external device, and which destroys
all the words that might describe the practice of delegation. It
also results in the double standard of physicists, who bring into
existence, more or less clandestinely, something that the mathematical structure of their theory officially denies: atoms interacting with fields, beams of electrons, beings that act, are acted
upon, and are affected, not merely measured. The "kitchen" of
approximations must resuscitate delegates and actors, whose
denial is the price of the "Hamiltonian miracle."
We are now able to understand the problem that confronted critics of quantum mechanics, who allowed themselves
to be seduced by the Schrodinger wave function, by the way it
extended the harmonious beauty of Hamiltonian physics. For
Schrodinger, quantum transitions had been the problem; for
these critics, the reduction of the wave function defined by the
Schrodinger equation is the problem. But in all cases, it is the
particular "narration" associated with experimental practice

THE

SILENT

DESCENDANT

THE

S R LE NT

DESCENDANT

97

that has disappeared, to the benefit of the intellectually satisfy-

situation of the frictionless fall for which they serve as a (good)

ing prospect of a general extension of the Hamiltonian repre-

approximation. Lagrangian, then Hamiltonian, dynamics have

sentation. However, the question is not whether this extension


is impossible. Some physicists, like David Bohm, have constructed mathematical beings capable of "reproducing" the
forecasts of quantum mechanics. Those beings were not taken
seriously because their construction came "later" and was
therefore incapable of challenging physicists' conviction that
quantum mechanics was a reliable guideup to and including the
"kitchen" to which it obligated themthat is, up to and including
the approximations it suggested. There is no doubt that Hilbert
space serves as a "lying factish," but its lie is very useful in the
sense that it has succeeded in suggesting procedures and relevant approximations, through which practice acquires meaning.
And it is this usefulness, this relevance of quantum formalism
as a guide rather than as a law, that is lost in the attempts to go
"beyond" the wave function. Naturally, "it is not impossible" to

developed a syntax that separates, with increasing radicality,


the power of presentation (and the mathematical definition
and invention that corresponds to this type of ideal situation)
and the "phenomenological" description of the systems we are
effectively confronted with. As we have seen, Hilbert space is
I he direct heir of dynamic inventiveness. The object of classical dynamics was constructed from the fixed point provided by
l he satisfaction of the causal equivalence between full cause and
total effect. But the construction of beings inhabiting Hilbert
space does far more than assert and presuppose the satisfaction
of a requirement that allows the problem of motion to be formulated and therefore still alludes to the physicist who formulates
that problem. This construction brings into existence beings
from whom nothing can be demanded, beings that cannot present a problem to anyone.

construct the mathematical expression of a "nonlocal" field that

Like causal requirements, Hilbertian requirements coin-

produces "localization" effects, but this expression is extremely

cide fully, in a way that is both necessary and sufficient, with

complicated. Gorrelatively, the relevant ways of "leaving Hilbert


space" are no longer presented as "plausible" on the basis of such
a field. In other words, the attempts to construct a "factish" that
does not lie, to the extent that they try to reconstruct the behavior of the wave function, confirm the requirements inherited
from the triumph of the Queen of Heaven and are committed to
further conceal the continuously negotiated character, fecund
but conditional, of the alliance between Hamiltonian formalism
and experimental practice.
Here, we confront a very unique situation whose uniqueness exacerbates, while transforming, the historical reciprocalcapture between "realism" and "causal measurement." From
inception that history has brought about a binary register of
evaluation. Galilean rolling balls satisfy the requirements of
causal equivalence only to the extent that they manifest the ideal

the exhaustive definition of the beings that satisfy them. That is


why those requirements are, at the same time and identically,
vectors of obligation, conferring on what they define the power
to engage, authorize, or prohibit. Unlike causal requirements,
however, Hilbertian requirements do not communicate with a
property of the object that may well require idealization but that,
nevertheless, makes intelligible a determinate experimental
practice. That is why, unlike the classical dynamic object, the
quantum object does not appoint a physicist-judge whose categories of interrogation would be those to which the object itself
is subjected. The measuring physicist appears, rather, as an
intruder. She breaks the superb mathematical symmetry of the
quantum object, asks questions for which no answer should be
able to be given as long as we "remain within Hilbert space."
We might object that, strictly speaking, the classical object

THE

SILENT

DESCENDANT

described in terms of cyclical variables no more accounts for the


possibility of its description than the quantum object described
by a superposition of eigenstates of the Hamiltonian operator.
Both are literally "unobservable" in the sense that any observation presumes an interaction and in that they are both defined
mathematically as devoid of interaction with an external world.
But the possibility of constructing a cyclical representation is
not something causal equivalence requires. This equivalence
enables us to establish dynamic equations but does not guar-

THE

SRLE N T

DESCENDANT

97

of phenomenological physics. Such equations all ask the same


question: "What will happen if . . . ?" On the other hand, Hilbert space provides answers that do not satisfy any question. It is
made to bring about pure affirmation, an affirmation that needs
"something else," the measurement that reduces the wave funclion, in order to become relative affirmation, relating Hilberlian beings to the device that interrogates them. In order to be
associated with a problem, Hilbert space must be violated.
It may be concluded that physicists' questions and misgiv-

antee that they can be integrated. Moreover, in the customaiy

ings do not refer primarily to any "grand ideas," to realism,

formulation, the dynamic system is defined as closed, with no

determinism, or any others, but to their own constructions,

place for the observer. But this definition acknowledges, with-

when read in terms of requirements and obligations. In no way

out dramatic consequences, that this closure is an idealization,

does quantum mechanics challenge the "existence" of a "real-

neglecting the interactions that enable observation. Such inter-

ity in itself." Rather, what is challenged is the relevance of the

actions can no longer be neglected in the quantum case, as Bohr

requirements of quantum mechanics when expressed in terms

untiringly emphasized. And Hilbert space, because it asserts the

of statements that bring quantum objects into being (let there

quantum analog of classical integrability, transforms into an

be a quantum system represented by a wave function \|/ . . .).

axiom the fact that time evolution cannot be related to interac-

But, in answer to this challenge, the definition of requirements

tions. As a result, the very fact that the quantum object cannot

that would restore the possibility of defining quantum objects

communicate with experimental practice becomes part of the

belongs to physical-mathematical inventiveness, and is relative

definition to which this object obligates us.7

to its historicity. There can be no question of physicists start-

There is a connection between the "practical lie" of quantum mechanics and the distinction between causal requirement, on the one hand, and the Hilbertian mode of existence,
on the other. The causal requirement concerns the way of defining a problem that corresponds to a situation characterized by
measurable variables, and the way the equations expressing that
problem are written, whereas, to follow the analogy, the Hilbertian mode of existence doesn't correspond to any problem. This
distinction matters because the practice of physicists, classical
or quantum, requires the mathematical definition of a problem, one corresponding to the situation they create. This practice is no more excluded by the equations of classical dynamics
than it is by kinetic equations, for example, or by the equations

ing from scratch, of contemplating the world with fresh eyes,


of escaping a history that does not belong to physics alone. Not
only are all physical entities, without which our experimental devices would have no meaning, part of that history, but, in
their own way, and even if it's in a way that is slightly deceptive,
a large number of the instruments that populate industrial or
hospital labs, as well as many commercial products, work hand
in glove with Hilbert space and confirm its "relevance," given
the approximations that "concoct" it, that force it to have consequences it is unable to justify. Is it possible to negotiate the
meaning of this relevance without having to introduce the violation of the lying factish?
But why introduce the problem? Maybe because "grand

THE

SILENT

THE

DESCENDANT

ideas" do not lack for importance in the history of mankind.


From Schrddinger's cry, claiming that his cat has to be able to
die even if no one can observe it, to Vigier's fury, from Einstein's
protest to Popper's plea for a "realism of propensities," from
Bohr's obsession with Einstein's objections, which troubled
him until the time of his death, to Nancy Cartwright's accusation
of lyingall of this, it seems to me, represents the expression of
a profound dissatisfaction, the feeling that history cannot stop
there, left with the irony of a physical-mathematical language
whose requirements turn against the intelligibility for whose
sake they were conceived.
Can we "civilize" the strange quantum descendant of the
Queen of Heaven, bring it to accept practices it could not exist
without, bring it to utter words that haven't been coerced? Can
the Queen of Heaven herself engender ideas of intelligibility
other than those her Hilbertian descendant has so paradoxically
satisfied? If this exploration concludes with such questions, it
is not accidental. All narration is also always a suggestion, with
the twofold risk of captivating those it addresses, or enabling
them to condemn the biases of the one who produces it. To
explore the question of quantum mechanics presents nearly as
many problems as choosing a path and viewpoints for exploring
a labyrinth, where paths continuously multiply and bifurcate.
Without a compass, losing our way is all but guaranteed, just as
following a path that leads to a quantum justification of parapsychology, or the strange doctrine of multiple worlds, or the
(strong) anthropic principle will lead us astray. The rejection
of dizzying perspectives (see Cosmopolitics, Book I) is, however,
far from sufficient for explaining the way in which I have presented quantum mechanics and its dissatisfactions. To identify
a source of dissatisfaction is never innocent, and the one who
does so often conceals in her sleeve the hypothesis of a solution
to the problem identified.
I don't intend to make a mystery of what guides me. It is

SRL E NT

DESCENDANT

97

loward the perspective that situates me that I now turn, toward


a response to the challenge of the Queen of Heaven and her Hilbertian descendant, unique in that it experiences no nostalgia
I'or the lost intelligibility of dynamics. From the point of view
defended by Ilya Prigogine, dynamics and quantum mechanics
are both equally unsatisfactoiy, for one reduces the difference
between past and future to the imperfection of our understand i ng, and the other to the act of measurement, or the act of awareness. Both condemn any possibility of coherence with the body
of other practices, scientific, technological, or cognitive, all of
which assume the nonequivalence of past and future, the "arrow
of time."
The uniqueness of Prigogine's position stems from the fact
that he felt obligated as a physicist by the need for consistency
between physics and these other practices. But, ultimately, this
obligation forced him to "swim upstream" against the current of
history, to that moment in physics when the Queen of Heaven
asserted her triumph and made kinetic description an "incomplete theory of dynamics." This is the meaning of the manipulative suggestion I have employed on the reader, the suggestion to
allow the moment of perplexity experienced at the conclusion of
Book III to extend to the dissatisfaction I, in my own way, have
associated with quantum mechanics. There is no answer without a question, and no question that does not translate a choice
that precedes it, and whose terms it cannot by itself justify. In
this case, the "manipulation" is nothing other than the translation of what has, since I began to take an interest in it, activated
my questions about physics.

BOOK V

In the Name of
the Arrow of Time
PRIGOGINE's

CHALLENGE

The Arrow of Time

The expression "the arrow of time" does not cause any problem
lor physicists, but casts some critical philosophers into an abyss
of perplexity. It is as if the physicist had taken it upon herself
lo speak about time "in itself," whereas her time is, and can
only be, the time found on a clock or any other device capable
of providing a quantitative measurement of change. Time is the
number of motion, Aristotle said. From this critical viewpoint,
physicists' time has the status of a condition of knowledge. It
enables us to study motion but should not be confused with it.
Yet, doesn't the fact of discussing the arrow of time manifest this
confusion between "time-as-measurement" and "measured
motion"?
The critics are obviously correct. The arrow of time is a metaphor. But they are mistaken if they believe that condemning
this metaphor will get them veiy far, for behind the metaphor
can be found the problem it points to, a problem criticism cannot make disappear.
Time, as the number of motion, is a general statement that
applies to all motions. Usingthe same chronometer, I can number the movement of a tortoise or a running horse, a ball or the
decay of a corpse. In that sense, it is one "external" measurement
among others. Readers may recall (see Book II) how "work"

110

T H E A R R O W OF T I M E ILL

T H E A R R O W OF T I M E

was defined by nineteenth-centuiy engineers, who indiffer-

a speculative statement in questionable taste but an illustration

ently measured the activity of rats, men, or the rolling ball, with

of an essential property of the equations of dynamic evolution,

or without friction. Just as, unlike rats and workers, the ideal

that is, the strict equivalence between two mathematical trans-

Galilean ball provides work with an intrinsic theoretical mean-

formations: the reversal at a given moment of the sign of all

ing, that same ball binds the relationships between "time" and

velocities in the system and the reversal of the sign of the time

"motion" that will create the problem of the arrow of time. The

parameter. This is what the perfect pendulum realizes, right

beat of the pendulum not only creates the numbers that will be

before our eyes, whenever its velocity changes direction at the

used for the measurement of all other motions, it is itself sub-

end of a cycle. If we consider this end point in the "present,"

ject to a temporal law that explains and guarantees the standard

the description of the future motion of the "descent" of the per-

it constitutes. In other words, the regularity of pendular motion

fect pendulum will not be distinct from that which would lead

is not only a condition of measurement (the way the apparent

it "back to its past," that is, from a description that led it to the

motion of the sun once was). Of course, it does not directly tell

present when transformed by a reversal of the time parameter.

us about time. But its telling concerns the articulation of suc-

To say that the pendulum goes "back to the past" can hardly

cessive instants of motion, which are subject, for the ideal pen-

be taken literally since nothing in the behavior of the pendulum

dulum as for any other dynamic motion, to causal equivalence.

can serve as a witness to its having accomplished this great deed.

The instantaneous state is then able to define past and future

In fact, the veiy definition of the ideal pendulum is that it does

states, which are all seen as equivalent. Correlatively, the beat

not "speak," in whatever sense of the word we choose, no matter

of the pendulum not only gives motion a number, the number

how metaphoric. The ideal pendulum entertains no relationship

itself becomes a function (of gravitational acceleration and the

to the world that could be understood as bearing witness. The

length of the wire). Time-as-measurement has become internal

ability to speak of going "back to the past" is, in effect, strictly

to the object. It enters into the definition of the object, and this

equivalent to the claim that its oscillating motion is fully deter-

definition, in turn, determines the existence of the object in its

mined by the conservation of cause in effect. In other words,

relationship to time.

We see that the metaphoric possibility of shifting from


dynamic evolution over time to time itself, as understood by
dynamic evolution, points veiy precisely to the singularity of the
dynamic object, which causes description and reason to coin-

the motion of the ideal pendulum, precisely to the extent that it


can challenge the difference between past and future, can cause
nothing but itself. It cannot "tell time" to anyone because it is in
relation to nothing.
The example of the pendulum reflects, in nearly paradigmatic

cide. The question of the arrow of time belongs to the field of

fashion, the power relationship between two claims of "time-

mathematical physics and is, in that sense, distinct from philo-

as-measurement": the time the pendulum gives expression to,

sophical questions such as "What is time?" It asks the question

the standard of measurement, and "dynamic time," introduced

of the meaning (and scope) of dynamic symmetiy as illustrated

by the equations of dynamics. In order for the pendulum to "tell

by the motion of a frictionless pendulum or the Galilean ball

time," it must be supplied with an external observer who "sees"

returning to its initial height. Similarly, whenever the literature

the pendulum, who counts the number of beats, who uses it as

of physics refers to a phenomenon "going back in time," it is not

a measurement standardactivities that cannot be reduced to

110

T H E A R R O W OF

T H E A R R O W OF T I M E ILL

TIME

dynamic evolutions. By means of the observer who counts the

of measurement presented by von Neumann. For von Neu-

beats of the pendulum or, armed with a chronometer, measures

mann, there is apparently nothing that authorizes the physicist

its period, as she might measure the movements of a horse or the

to "stop" quantum description, in terms of the wave function,

gestures of a laborer, the human practice of measurement asserts

when interacting with a measurement device. But if he rep-

the difference between past and future. But the dynamic equa-

resents the device in quantum terms, in Hilbert space, he is

tions for pendular motion introduce the self-determination of

obligated by that representation to recognize that he has lost

the motion of oscillation, the autonomous variation of positions

the ability to characterize a measurement device. At this point,

and velocities that allows it to be characterized exhaustively as

the human observer, endowed with consciousness, is required

an existent in time, in a way that denies the difference between

by von Neumann to defrne the conditions that make quantum

past and future.


We might conclude that the two modes of existencethat
of the observer and that of the pendulumcorrespond to two
meanings, external and internal, of time-as-measurement. The
correlate of this conclusion would be that the mode of existence
of the subject who observes obviously has nothing to do with
physics and cannot, therefore, present a problem for the physicist. However, like any conclusion that introduces a "subject"
confronting the "world," this one gets ahead of itself. It lends
itself perfectly well to "high-level" meetings between philosophers and physicists, and irresolvable confrontations where
eveiy participant claims to represent an authority that has the
power to make determinations for the others. Yet it overlooks
the fact that there is no need for a human "subject" for the pendulum to be able to "tell" the time or to bear witness. Any clock
will do, and can be integrated, as is the case in any physics labo-

measurement possible. But as we saw in discussing the work of


Nancy Cartwright, this approach to measurement, which creates
a dramatic conflict between human awareness and "Hilbertian
reality," is remarkably artificial. Just as the clock, if it is to function and tell time, must break the ideal symmetry of dynamics,
the quantum description of any process where time "counts" for
physicists obligates them to 'leave Hilbert space"which they
do so unhesitatinglythat is, to introduce the approximations
that will free them from the obligations imposed by the wave
function. In both cases we are dealing not with a confrontation
between "mankind" and its "object" but with a question produced by physics. It is physicists, not human beings in general,
who find themselves facing conflicting obligations: between
their obligations vis-a-vis relevant experimental practices and
their obligations vis-a-vis a Hilbertian or dynamic definition of
"reality."

ratory, into an automated device that "traces" the evolution of a

The parallel between the question of the arrow of time and

property measured over the course of time. The question of the

the problem of quantum measurement extends even further.

arrow of time arises, in fact, within physics, for example, when


questions arise about how to describe the (dissipative) interactions that are required to connect the pendulum to the hands of
8

a clock. Do the laws to which the pendulum is subject reveal the


truth about the clock?
It won't escape the attention of readers, or anyone familiar
with quantum mechanics, that we are very close to the problem

One way of saying that the clock must violate the theoretical
ideal that claims to describe it is to recall that its movement will
proceed toward equilibrium if it is not maintained. Similarly, as
we saw in Book IV, the interaction of measurement can be compared to an irreversible evolution toward equilibrium, which is
translated by the production of the permanent "mark" without
which an experimental datum has no meaning. In both cases the

110

T H E A R R O W OF

T H E A R R O W OF T I M E

TIME

infraction can then be considered an approximation. Because of

ILL

extension, all humansmust accept that the world we charac-

its dissipative interaction with the gear mechanism, the clock's

terize as yielding to observation is, first and foremost, defined

movement approaches, but cannot achieve, the ideal of the

by its distance from intelligible reality, dynamic or Hilbertian,

pendulum. Similarly, the evolution toward equilibrium that is

of which it is merely an approximate translation. On the other

required by quantum measurement could be compared to the

band, at work the physicist requires only that theoiy be relevant.

"forgetting" of certain interference terms, the approximation

Whenever it is a question of preparing and predicting observ-

corresponding to this forgetting being justified by the fact that it

able processes and properties (lifetime, cross section, and so

has no observable consequence for the measured datum.

on) that imply a difference between past and future, the physi-

That is why in the quantum as well as the classical case the


need to "violate the law," to break the symmetiy of the fundamental equations, far from seeming to challenge the authority of
the law, can very easily be concealed. In fact, the approximation

cist "negotiates," without the least qualm, the transition between


"fundamental" theoiy and the relevant representation, and uses
this successful negotiation as an experimental confirmation of
t he theoiy.

seems to confirm the authority of the law. Even more so since

This also results in the double status of the meaning assigned

it can mask itself behind a perfectly anodyne appearance, and

to observable "dissipative" properties. Dissipation is both

appear fully justified by the insignificance of what is being over-

required, because it characterizes the majority of experimen-

looked. In fact, once the "violation" can be defined as negligible,

tal situations that legitimize theoiy, and disqualified, because

the game is over. Just as the most negligible "kitchen concoc-

it is considered relative to the approximation resulting in the

tion" is sufficient to enable one to leave Hilbert space and end

construction of their theoretical meaning. Dissipative proper-

up with a probabilistic description that is realist in nature, the

ties are thus included for, if they did not, the theoretical scope

least approximation of the defined ideal suffices to introduce

of theoiy would be restricted to a handful of special cases and

the arrow of time into a description governed by dynamics.

the theoiy could not claim to be "fundamental." But the fact that

More accurately, in both cases, any approximation always results

they conflict with theoretical syntax is not taken into account.

in breaking the symmetiy that singularizes the ideal represen-

The double standard by which physics manages the "conflict

tation. Consequently, constructing the meaning of what experi-

among obligations" that inhabits it is characterized by the pro-

mental practice obligates us to do is trivially simple.

motiondepending on the situationof two divergent obli-

The "double standard" that Nancy Cartwright discovered


in quantum mechanics is, therefore, a shared feature of both
classical and quantum mechanics. In both cases, the physicist
at work bears no resemblance to the physicist as she presents
herself to the philosopher or as she presents herself to herself
when attempting to reflect on the obligations of her science. In
such situations, the obligation attached to the ideal representa-

gations. The first is the obligation of experimental relevance:


theoiy does indeed account for what we observe. The second is
the theoretical obligation, more specifically, a reference to the
betrayal of this obligation. It would be only if physicists do not
take the obligations of theoiy as far as they can go that they may
be tempted to assign a physical meaning to observable "dissipative" properties.

tion takes precedence. Like measurement, the arrow of time is

This "double standard" defines the problematic space

merely a well-constructed fiction. "We"the physicist and, by

wherein the question of the arrow of time actually takes place

110

THE ARROW

OF

T H E A R R O W OF T I M E ILL

TIME

today. Anyone who attempts to argue that the way phenomena

be sufficient to condemn the vicious circle entailed by a judg-

bear witness to a "dissipative" temporality should be understood

ment that attributes the responsibility for the operation of time

as a source of obligation runs up against the fact that, currently,

measurement presumed by every description to the imperfect

these phenomena are not, in themselves, likely to challenge the

nature of that description. But the requirements of logic must

relevance of the "fundamental" equations because it is on the

bow to the facts. Physics is a history, and the fact is that at the

basis of those equations that they are described. That is why the

beginning of the twentieth century, that history experienced a

burden of proof falls entirely on the challenger. She must prove

genuine form of reciprocal capture that produced the respec-

that the means she proposes so that theory is able to compre-

tive identities of what we refer to as dissipative phenomena and

hend this temporality are not the by-product of a betrayal of the

level of fundamental description. While the description of phe-

obligations imposed by that theory, that they cannot be reduced

nomena, interactions, and dissipative events was considered an

to the surreptitious introduction of an approximation into the

"incomplete" description from the point of view of theory, the-

description, the arbitrary rejection of a possibility of measure-

ory was given a power of extension that nothing should be able

ment or discrimination provided for by that description. If she

to limit.

cannot, she will have confirmed that, in effect, the theoretical

In asserting the inseparability of the wave function and its

account of the dissipative phenomenon she proposed is relative

reduction, the Copenhagen interpretation attempted to foster

to the approximations that allow it to be constructed.

a solution that satisfred logic. The quantum being could not be

There is nothing logical or necessary about this challenge.

described, would have no assignable identity, independently

Quite the contrary. Logic, had it been allowed to function as

of what conditioned experimental access to it. However, logic

such, as a constraint with which the judgments of physicists

and the critical philosophers that relied on it are the only ones

had to comply, would suffice to tip the balance of judgment in

fully satisfied with this solution. Yes, it creates perspectives that

the opposite direction. It would lead us to describe quite dif-

can be described in grandiose terms. Heisenberg spoke of the

ferently the conflict among the obligations that inhabit physics.

person who, examining the sand on a beach, studies the tracks

The obligation of experimental relevance marks the distinction

whose origins he questions, finally realizing that they are his

between a physical theory and a revelation that appears out of

own footprints. Here, the observer is the only rationale for what

nowhere. Theory, therefore, is not logically empowered to dis-

he observes, which means reducing the damp sand to general

qualify, because of its own obligations, what is the very source

properties that may be taken for granted. The world, like the

of its authority. To return to the question of the relationship

sand capable of receiving an impression, is essentially silent.

between the pendulum and the clock, if the clock is the mea-

The answers we obtain are determined by the questions we have

surement device by which we were able to assign a law to the

decided to ask. This expresses the pathos of a situation wherein

motion of the pendulum, that law cannot claim to reduce the

physics, in the very act by which it claims for its equations what

dissipative nature of collisions and frictional forces to a simple

temporal symmetry warrantsan intelligible perfection from

approximate description, for it is because of those collisions and

which nothing escapesunderstands that those equations pre-

frictional forces that the pendulum is connected to the device

vent it from assigning any intelligibility to the world it thought it

that moves the hands and becomes part of a clock. Logic would

was addressing.

110

T H E A R R O W OF

TIME

No working physicist will ever fully acknowledge this criti-

T H E A R R O W OF T I M E ILL

silent world to which a ventriloquist physics would lend its own

cal lesson. Physicists accept the idea that we may not be able to

language. The histoiy of physics does not put us in a position to

conceive reality independently of what makes it observable. But

defend what it has set aside because its "decisions" are insepa-

is it necessaiyto continue the parallel that serves as my g u i d e -

rable from the creation of new problems and new beings. Just

to assign to the collisions and dissipative shocks that make time

as the histoiy of physics shifts from "dissipation" to "incom-

measurable the noble but opaque status of a "condition" for

plete description" via "reciprocal capture," it is clear that it did

there to be an "access" without which the ideal motion of the

not take the path Bohr suggested. More specifically, it bowed "in

pendulum would have no observable meaning? Is it wrong to

passing" to the inevitable character of those propositions, only

believe that those collisions and frictions are physical processes

to quickly sidestep them. Today, Hilbert space is presented as

to which we certainly delegate the role of "connector" between

the worthy heir of Hamiltonian dynamic space, that is, as sepa-

pendulum and clock hands, but which happen "regardless," the

rate from the "reduction of the wave function" that was suppos-

way an excited atom emits light regardless of whether this emis-

edly inseparable from it. Physicists "inhabit" that space, ask it

sion is a condition for assigning stationary states to the atom?

questions, use it to introduce problems "as i f " it supplied the

In fact, Niels Bohr, was no different than other physicists


in believing that the quantum world was not mute. But he was
haunted by the need to assert the novelty of that world, its irreducibility to the phenomenal world concerning which the arguments of mechanics are constructed. And his way of asserting
it was to insist that we could never dream of "inhabiting" that

reasons defining the quantum world. The veiy fact of introducing "approximations" into a Hilbertian formulation reflects the
fact that this formulation has been recognized as a description.
Only a description introduces the problem of what will be kept
and what will be ignored. In no case does a divinatoiy apparatus
authorize such a choice.

world, of constructing a "rational" account of it, identifying

Are we to believe that this survival of reciprocal capture

the intrinsic reasons explaining what we observe, the way we

reveals an irresistible psychological "penchant" on the part

believed we could construct the reasons explaining the Gali-

of physicists? Does the defeat of Niels Bohr, like that of Pierre

lean world. The quantum world will never reveal its "reasons";

Duhem before him, force us to treat the physicist's "faith" in the

in fact, it forces us to recognize the idealization behind the idea

possibility of a "vision of the physical world" as an inevitable

that the world of mechanical phenomena was able to assert its

ingredient of her practice? That would be somewhat prema-

own reasons. That is why, in my discussion of Bohr's position, I

ture. For histoiy can be expressed quite differently: opportu-

introduced the concept of the "divinatoiy factish." "Divinatoiy"

nity makes the thief. Let's return one last time to the analogy

here should be understood literally. Even though the physicist

between the quantum question and the relationship between the

constructs the message according to her own questions, it is

pendulum and the clock. If it is possible, at the cost of plausible

on signs that have arrived from another world, and not on some

approximations, to derive the description of collisions and fric-

amorphous noise onto which she would project her own mean-

tions from the theoretical understanding of the perfect pendu-

ings, that the physicist acts.

lum, why shouldn't we do so? If Hilbertian reality as such is able,

Yet, it is pointless to defend Bohr's proposition, to acknowledge that it has nothing to do with the impoverished idea of a

at the cost of plausible approximations, to account for the theoretical meaning of observable properties and processes that are

110

THE ARROW

OF

TIME

not part of the divinatoiy economy of quantum formalism, why


deny it what that ability allows it to claim? Why deny that this
Hilbertian reality has the right, as such, to obligate, to define
what the physicist deals with? In other words, although there is
indeed a question of "decision making" in the history of phys-

T H E A R R O W OF

TIME

ILL

other words, the mandate of whoever wishes to "reenact" the


question of the arrow of time, the status of the asymmetry characterizing "dissipative" processes, requires the invention of
means to demonstrate that the "progress of physics" entails that
these questions be taken into account.

ics, this does not refer us either to a subject and its penchants or

It can no longer be a question of unilaterally, that is, in an

to a onetime event but to a process of practical invention. What

ad hoc manner, adding an opaque dissipative factor to dynam-

"makes" the decision is the effective possibility of creating rel-

ics, which would ensure asymmetric temporal behavior when-

evant extensions to reciprocal capture, which stretch the ability

ever experimentally required, simply to resolve the problem. To

of the theoretical formulation to guide and interpret experi-

accept this addition "out of nowhere" would imply that physi-

mental exploration as far as possible. Like the "thief" produced

cists consider their strategy of approximation to be a bitter set-

by "opportunity," the physicist is primarily the product of this

back that must be corrected at all costs and by any means, even

process. It is not her psychological expectations that are satis-

ad hoc. In fact, the very opposite is true. The challenge is to turn

fied, however. She is the spokesperson of the satisfaction pro-

a situation generally felt to be one of the triumphs of physics into

duced by the operation of extension.

something unsatisfactory.

There is no a priori requirement such that the object of


a "decision" in the history of physics cannot be reenacted by
the history of physics. Nothing requires it, however, and what
I have just advanced allows us to predict that the opposition to
such a move will be stiff. For the physicist-actors in this story
are anything but impartial arbitrators. Of course, the nonphysicist might be sensitive to the claim that neither Hamiltonian
dynamics nor quantum physics has a "logical" right to equate
the asymmetry between past and present that characterizes dissipative processes with the breakdown of symmetry in their own
equations, or to judge this breakdown of symmetry in terms of
an approximation. The same argument is either dismissed by
physicists or, at best, receives an indifferent "yes, I know, but
all the same." I know that, in terms of logic, you are correct, but
the fact is that the symmetry-breaking approximations have
effectively allowed for the characterization of a very satisfactory ensemble of "dissipative" processes. As Einstein said about
Mach, logical criticism has never been useful except to kill the
vermin that live off physics, never to promote its progress. In

The decisions made throughout the history of physics cannot be redone in the name of right or logic, but require a process of invention that, in order to alter history, must become
part of the history being altered. And what has to be invented
is none other than an appeals procedure, one that provides de
facto ratification of reciprocal capture but challenges its modalities. In order to have a chance to convince physicists, this process must begin from a situation they judge satisfactory, one
that refers to dynamics and quantum mechanics as providing a
"complete theory," which rightfully holds the key to dissipative
processes. Correlatively, the question of dissipation must be
framed in terms of the breaking of the symmetry of the equations
of dynamics and Hilbert space. It is then, and only then, that the
judgment associating this breaking of symmetry with approximation can be appealed. In other words, the "reasons" justifying an arrow of time that cannot be reduced to an approximation
must be constructed in such a way that the symmetric equations
themselves supply the justification for breaking the symmetry
they assert. Naturally, such reasons must be recognized, along

110

T H E A R R O W OF

TIME

T H E A R R O W OF

TIME

ill

with those that are made explicit by the equations, as "objec-

I hat the future verdict will be fair." Today what has truly won

tive reasons" stemming from the object. For the proposed solu

is indifferencephysicists have chosen not to care. Strangely

tion cannot appear as a simple version that would differ only

enough, a new concept is now taken for granted, that of quan-

rhetorically from what physicists are already doing when they

tum "decoherence." This states that the slightest perturbation

introduce relevant "approximations." It is the phenomenon

to a quantum system is sufficient to destroy the superposition

itself, as represented in dynamic or Hilbertian terms, that musl

that characterizes ita serious concern for those who envis-

be recognized as imposing the obligation that the time symmetry

age the possibility of a quantum computer, and an easy way to

of the equations that represent it be broken, that must disallow

ascribe quantum "histories" to a universe defined in terms of a

the assertion that its time-broken symmetry can be explained in

wave function. This concept, however, is merely a name for what

terms of an approximation. Finallyalthough physicists might

physicists already knew, which is that the slightest approxi-

be divided about the need for this last conditionit would be

mation was sufficient to obtain a "realist" characterization of

good if the recognition of the specific obligations to which the

quantum actors. It may be a slap in the face to the generations

arrow of time corresponds were accompanied by the identifi-

of philosophers who meditated upon the "measurement prob-

cation of new experimental possibilities. Such identification

lem" but it also shows that physicists are pragmatic. If there is a

would endorse the superior relevance of the proposed solution

"true" need, they accept what they previously rejected, and the

in everyone's eyes. Progress could then resume its normal and

principles they proclaimed in order to justify that rejection are

moral course: as usual, the scene will be one in which nature

put to rest. It may well be that Prigogine's efforts will be associ-

itself distinguishes among its interpreters and narration can

ated with the memory of a useless, and desperately complicated,

solidify into a tale of progress that disqualifies error.

project and the misplaced ambition to transform decoherence

It is at this point that the narrative choices for my stories


need to disclose their rationale. For the requirements I have just
described impersonally are the very same that were discovered,
constructed, and accepted by Ilya Prigogine. 3 Can we say that
these requirements have been honored and that the statements

into a matter of principle, which is now known to be a property


of quantum systems. What is certain is that texts such as this will
not reverse the course of events. For physicists, too many things
are at play for an "incompetent" intervention to be considered
anything more than presumptuous chatter.

resulting from Prigogine's work will be confirmed by his col-

However, the situation is different from the point of view of

leagues? In the French edition of Cosmopolitics, I wrote: "No one

the questions that guide my exploration of physics, those that

can answer this question. Will the existence of doubt and oppo-

deal with the consistency of our practices of knowledge. Here,

sition one day be considered 'merely sociological' in the sense

the only thing that counts is the fact that hesitation was pos-

that it would cast light on the 'habits' of physicists, the 'pres-

sible. In light of the requirements I have just detailed, we can

tige' of physical laws, the 'ideal' of perfection associated with

understand how a potential verdict that claimed these require-

those laws, or the overwhelming 'authority' of the principles of

ments had been satisfied would have celebrated the coming into

symmetiy? Or, rather, will Prigogine's effort be seen as sterile,

existence of a new physical-mathematical factish, able to claim

futile, or 'unfortunately condemned by some hidden defect'?"

that it spells out "nature's reasons" better than its predecessors.

"History," I wrote, "hesitates, without the slightest assurance

By means of this verdict nature could be considered to have

T H E A R R O W OF

110

T H E A R R O W OF

TIME

TIME

ill

appointed a suitable representative. However, the very possibil-

to make what might be called the "Galilean miracle" an active

ity of hesitation introduces the question of what we can demand

protagonist in this history. Physicists who are heirs to that mira-

of phenomena that we represent and the obligations associated

cle may maintain that they are authorized to claim that the prob-

with the ambition to represent them. Of course, it is not a matter

lem they formulate about reality is true to the "problem" that

of challenging the importance of the verdict, the importance of

reality itself continuously produces and resolves from moment

the invention of a new definition of "nature's reasons." Rather, it

to moment. In that sense, what physicists call the "fundamental

is a question of emphasizing that the invention of this new defi-

laws" (of physics or nature) support their claim to be the spokes-

nition is precisely what was required by physicists who inher-

persons of reality. But the achievement those laws bear witness to

ited the reciprocal capture that marks the history of physics.

does not transcend the singularity of the tradition that has pro-

The notion of "nature's reasons" itself does not refer to a quest

duced such a claim. For the possibility of maintaining it serves

that would transcend the practices of physicists. It is inherent

as a defining condition, selecting which features of reality can

in the requirements imposed upon those who suggest that they

have a legitimate spokesperson. Indeed, this is the miracle

abandon their double standard: they must be forced to do so by

that Prigogine has been called upon to repeat: the construction

"nature itself."

of a physical-mathematical representation that must be able

And the challenges are great. For the physical-mathematical

to claim the strange ability to "bring into existence" a reality

beings that might claim the ability to impose obligations that

endowed with the power to dictate its own (and new) conditions

entail the need to take dissipative phenomena into account will

of representation, and thus capable of maintaining the claim

not be met with sympathy, as the vectors of a long hoped-for

that it transcends all the "realities" with which our other prac-

solution. They will have to overcome a certain amount of skepti-

tices engage. That physicists can require the repetition of this

cism. Questions will arise about whether the claim with which

"miracle" as a condition for taking the arrow of time seriously

they identify hasn't been extorted from them; about whether

reflects the fact that this pretense to transcendence, a vector of

the obligation they bring into existence does not in fact imply

fascination, is not legitimized by their achievements because it

that they forfeit reference to an exact definition, one that is pos-

follows from a definition of achievement that is strictly inherent

sible in principle. And they will be asked if they are capable of

in their tradition.

"keeping up," of having the behavior expected in all situations

Some may consider the following narrative to be dry, oth-

in which they might appear. Can they match the performance of

ers will fmd it stimulating, and there are those who may even

their Hamiltonian rival whenever the conditions of the encoun-

find it offensive. To trace the "technical" invention of problems

ter have been defined? If they fail to satisfy all of these require-

and instruments is not to present, as I have done with Hamilto-

ments, they cannot hope to gain recognition as representative


of an "autonomous reality" that itself stipulates how it should be
described.
Whatever verdict history reserves for Prigogine's proposal,
its interest will remain to confer on physics an irreducible, historicalthat is, path-dependentdimension, more specifically,

nian dynamics or Hilbert space, a stable physical-mathematical


"being," whose behavior we wish to grasp. If Prigogine had been
successful in his endeavor, the last step of this narrative alone
would survive, which would supply the premises for a functional
presentation of the new being. Everything else would be relegated to the status of a scaffolding that can be forgotten once

THE ARROW

OF

TIME

construction is complete, with some elements being redefined


as applications of the new theoiy, interesting only for those who
wish to use them within a specific field. The history staged by my

narrative has behaved in just this way: each of its steps has redefined the problem by making the shortcomings of the previous
step explicit. In this way, each step incorporates the preceding

Boltzmann's Successor

step, but also allows it to be disqualified, to show how it was


unable to provide a satisfactoiy solution to the (new) problem.
Consequently, this narrative will be diy for those who perceive
no more than an incomprehensible labyrinth, populated with
strange beings who suddenly appear grimacing before us, only
to disappear at once, as in a haunted house in an amusement
park. However, it will be stimulating for those who recognize the
passion that split this labyrinth, the tireless effort that created
these beings, and hoped to see them fulfill their intended task,

"To us the only hope for obtaining a general theoiy of non-equilibrium processes seems to be to reformulate the entire prob-

but was prepared to destroy them and incorporate some of the

lem in a more systematic way on a purely mechanical basis." 1

attributes they had managed to sustain in new beings, contain-

The primary function of this quote, which dates back to 1963,

ing new ingredients, let loose across a modified problematic


landscape. Finally, it will certainly be offensive for those who
maintain, for one reason or another, that a pipe dream produced what tried to come into existence but was destined for
failure, and see in its meanderings proof that, in any event, the
path would lead nowhere.

is to dissipate a recurrent misunderstanding. We should not


understand "general theoiy of non-equilibrium processes" to
mean the so-called theoiy of "dissipative structures" in the field
of "far-from-equilibrium" thermodynamics, for which Prigogine later became famous. Often, Prigogine's work on the "arrow
of time" is treated as an extension of his work on thermodynamics. But the misunderstanding is not entirely innocent, for it
is used to level the accusation that he suffered from "delusions
of grandeur." After Prigogine had shown that irreversible processes, which produce entropy, can, at the macroscopic scale,
be the origin of new regimes of activity that are impossible at or
near equilibrium, it seemed as if he had convinced himself that
no one should be allowed to ignore irreversibility and dissipation. Supposedly, this is the reason he embarked on his overly
ambitious attack on dynamics and quantum mechanics. Naturally, such statements are often accompanied by a slight sneer.
Athermodynamicist has no business "playing with the big boys"
or attacking the "fundamental problems of physics."
133

124-

BOLTZMANN'S

SUCCESSOR

In fact, the man who, in 1962, revealed his ambition to construct a "general theory of non-equilibrium processes" did not
do so as a specialist in thermodynamics, the heir to Lagrange,
Carnot, Clausius, and Duhem, but in statistical mechanics, and
as an heir to Lagrange, Hamilton, Maxwell, Boltzmann, and
Gibbs. And he articulated the problem at the exact moment
when dissipative physical-chemical processes were being interpreted in terms of dynamics: when Boltzmann's initial ambition
of constructing a theory for the irreversible approach to equilib rium was transformed into an interpretation that simply eliminated the problem presented by irreversibility.
In Book III I described how Boltzmann had constructed
what he hoped would be a "purely mechanical" representation
of the simplest nonequilibrium process, one in which a dilute
gas reaches thermal equilibrium. The gas is then represented
as a population of elastic particles whose respective velocities change because of collisions. Boltzmann ended up with an
"integral-differential" equation that represented the manner in
which the collisions cause the statistical distribution of velocities in the population to evolve over time. Through this equation he was able to construct his well-known r^-theorem, which
he hoped would provide a rigorous mechanical interpretation
of the growth of entropy. As we saw earlier, his interpretation
was unable to resist the paradoxes of Loschmidt and Zermelo,
namely, the obligations imposed by the Hamiltonian trajectories that Loschmidt and Zermelo made explicit for the first
time. Boltzmann was forced to recognize that his theorem did
not describe the impossibility of an evolution that would lead to a
spontaneous decrease in entropy, thereby contravening the second law of thermodynamics, but only its improbability.

BOLTZMANN'S

SUCCESSOR

the expectation that a purely mechanical derivation of the


growth of entropy could be found? It was deprived of any
authority, even though it was still useful for interpreting the
phenomenological properties of gases in kinetic terms. The collisions it introduced could no longer pretend to "explain" the
irreversible evolution of a gas toward equilibrium. Those collisions had appeared to Boltzmann to be responsible for that
irreversible evolution, but only because of the approximations
concealed in his reasoning. When the Queen of Heaven, the
Hamiltonian definition of dynamic trajectories, claims the full
scope of her powers, collisions are "smoothed" and are unable
to produce any effects other than those produced by the other
dynamic interactions.
Therefore, the assimilation of the kinetic description to an
incomplete theoiy of dynamics was made possible because of
a judgment about "kinetic collisions," as introduced by Boltzmann. Physicists concluded that such collisions, which were
used to construct a model for the growth of entropy, were merely
phenomenological, a simple approximation of "collisions in a
rigorous dynamic sense," that is, devoid of any possibility of giving meaning to the arrow of time.
Let's return to Prigogine's goal of 196?, the construction of a
"general theoiy of non-equilibrium processes," that is, a theoiy
that extends beyond the scope of validity of the Boltzmann equation (dilute gases). In the context I have just recalled, that goal
assumes a somewhat paradoxical guise. For, reduction of the
kinetic collision to a simple approximation of a dynamic collisionhas one apparently obvious consequence: the only "general"
theory possible must take as its subject collision in the dynamic
sense and, in doing so, transforms the description of nonequi-

Therefore, it is the association between probability and

librium processes into a form of dynamic evolution like any

entropy that was retained as the essential lesson from Boltz-

other. If Boltzmann's equation introduces a "mechanism" whose

mann's work. So what became of the evolution equation for

role derives from an approximation, its limitations should pres-

the distribution of velocities that had served as the basis for

ent no theoretical problem. We should not hope for, or demand,

BOLTZMANN'S

126

SUCCESSOR

BOLTZMANN'S

SUCCESSOR

127

a more general theoiy. It is normal that a physicist would limit

(yet) try to explore the obligations associated with irreversibility

herself to refining, on a case-by-case basis, the most relevant

where dynamic descriptions are involved, but only to construct

technique of approximation for a kinetic interpretation of the

the "profound physical meaning" of the obligations associated

phenomenological evolution she is studying.

with a relevant kinetic description of the various irreversible

And yet, in 196?, Prigogine and his colleagues held the

processes. What Boltzmann succeeded in doing with dilute

opposite position. Their argument was based on the experi-

gases, Prigogine wanted to do in other milieus, which, not being

mental relevance of the kinetic description: "There is certainly

dilute, require that interactions other than purely kinetic colli-

a profound physical meaning in the Boltzmann equation, as is

sions be taken into account.

borne out by its remarkable agreement with experiment in the

Prigogine, when considering the relative modesty of his ini-

calculation of transport coefficients for dilute gases."* It is this

tial ambition, often stated that he was fortunate that, at the time,

meaning that must be made explicit if it is to be extended to "the

he didn't "dare" try to imagine where those questions might

enormous range of experimental conditions in which trans-

lead. In fact, the terrain he explored was both reassuring and

port or relaxation phenomena are now being studied. Starting

crucial. Reassuring, because it limited the scope of the question

from low-temperature transport processes in liquid helium and

of irreversibility to the field of statistical mechanics alone, with-

mounting to high-temperature processes in fully ionized plas-

out asking the question of the latter's status. Crucial, because

mas, the range of energies covers ten powers of ten!" 3 The ambi-

the requirements of a general theoiy of irreversible processes

tion to address what remains "one of the fundamental problems

would stage the scene in such a way as to make explicit the obli-

of physics""time, so closely related to irreversibility"is for-

gations of such a theoiy. Those obligations would constitute a

mulated briefly, but there is no question (at this stage) of reviv-

stable referent, a "pole" serving as a compass during Prigogine's

ing the confrontation between the mechanical model of entropy,

quest. This would subsequently enable him to identify physical -

Boltzmann's

function, and the "attributes of the Queen of

mathematical developments that "could lead in the right direc-

Heaven."4 The problem is limited to the contrast between the

tion," which could be put to use in studying irreversibility. What

theoiy of equilibriumwhich is general and can be used to

some have seen as opportunism reflects this "on the watch"

define the equilibrium of any physical medium from liquid

position. From statistical mechanics to chaotic systems, and to

helium to plasmasand the great difficulty, once we leave the

today's "rigged Hilbert spaces," the point of view remains the

domain of dilute gases, of constructing a kinetic description of

same: irreversibility implies obligations. 5 What will continue to

the processes that lead to equilibrium.


In Book III I referred to the growth of entropy as an "enigmatic factish." Entropy, whose growth appears to characterize

evolve, however, are the status and scope of those obligations,


that is, the twofold determination of who those obligations will
affect and the objects likely to impose them.

irreversible change, in fact defines only the equilibrium state

So, in 1963 the program was limited by two stable refer-

that is the conclusion of that evolution. The increase in entropy

ences whose stability Prigogine did not challenge. One was the

itself does not deliver its physical meaning in thermodynamic

Boltzmann nonequilibrium equation, which represented the

terms. Could kinetics supply that meaning? This was Boltz-

(irreversible) evolution in time of the distribution of velocities

mann's hope and, in 1962;, it was Prigogine's as well. He didn't

under the influence of collisions, which is only valid for dilute

134 B O L T Z M A N N ' S

BOLTZMANN'S

SUCCESSOR

SUCCESSOR

135

gases. The other was the general theory of equilibrium in sta-

have in a dynamic description. The collisions that occur during

tistical mechanics, which represented the probability distribu-

(he abnormal evolution cannot, therefore, be predicted solely

tion, p. associated with Gibbs, and its definition at equilibrium,

on the basis of the velocity distribution function.

pi(j, which is constant over time. Let's begin with the Boltzmann
equation and see how its limits of validity are interpreted.
The existence of those limits were rather dramatically

It was to accommodate the singularity of this "abnormal"


evolution that physicists since Boltzmann have introduced "correlations" among particles. This evolution leads the system away

exhibited by Loschmidt's thought experiment with the reversal

from equilibrium because the collisions that produce it "repeat"

of velocities. When the direction of all the velocities of all the

the past. That is why Boltzmann's hypothesis that the positions

particles in a gas evolving toward equilibrium is instantaneously

of particles have no relevance for the evolution of the veloc-

reversed, the evolution resulting from the new initial state

ity distribution function, now known as the "molecular chaos

should, according to the equations of dynamics, lead the system

hypothesis," in this case ceases to be valid. Collisions are pro-

away from equilibrium just as spontaneously as it approached it

duced between correlated particles.

previously. Why wasn't this evolution predicted by Boltzmann's

Correlation is a physical-mathematical being liable to dis-

equation? Because Boltzmann's argument was based solely on

concert readers who seek to retain intact a distinction between

the distribution of velocities. His equation introduced types

what they accept as belongingto physical reality (forces of inter-

of collisions characterized by the velocities of particles as they

action, trajectories) and what we associate with the questions

"enter" into collision and as they "exit" those collisions, and

we ask and the instruments we invent to ask them. To which of

it assumed that the relative frequencies of the different types

these two categories do correlations belong?

of collisions depended solely on the distribution of velocities

If the solution to the problem were explicitly constructed, if

among the population of particles. In short, the fewer the num-

the variation over time of the position of each particle in the gas

ber of particles with velocity vt the fewer the number of colli-

were described explicitly, there would be no place for correla-

sions involving a particle with that velocity. This seems to be

tions, which translate the limits of approximation associated

common sense. However, Boltzmann's assumption, in order to

with the hypothesis of molecular chaos. Correlations, therefore,

be valid, assumes that the information about the positions of

do not appear able to lay claim to the "reality" of trajectories and

particles at a given moment has no relevance for calculating the

are even explicitly condemned by the way the integration of the

frequencies of the different types of collision at that moment.

equations of dynamics is defined. Recall that a system whose

Yet, after the velocities are reversed, the description of the gas

equations have been integrated can also be described in such a

as it moves away from equilibrium must make use of those posi-

way that it becomes isomorphic with an ensemble of indepen-

tions. The initial state of the evolution away from equilibrium

dently evolving "modes." In that case neither the concept of

implies the definition of the position and velocity of each parti-

collision, nor, a fortiori, the concept of correlation retains any

cle. Any small modification of the position of a particle will lead

meaning. The existence of correlations appears to be purely

the gas to quickly resume its "normal" behavior toward equilib-

relative to the fact that we don't know the trajectories, that is,

rium. In other words, the positions and velocities acquire equiv-

to questions that our ignorance and not the system itself obligates

alent roles in the description of the evolution, the very roles they

us to ask.

BOLTZMANN'S

SUCCESSOR

BOLTZMANN'S

SUCCESSOR

And yet correlations in statistical mechanics belong to a

gas toward equilibrium, providing that this relationship incor-

realist syntax. More specifically, their status is comparable to

porates the question of correlations "before" and "after" the col-

that of the collisions themselves, whose definition they com-

lision. Boltzmann had assumed that the position of the particles

plete. Boltzmann defined collisions in terms of the velocities

didn't matter. Loschmidt created a situation in which, as in

of particles that "enter" into collisions and by the velocities of

dynamics, they play a role equivalent to velocities. The question

particles that "exit" collisions. The complete kinetic defini-

of correlations transforms the confrontation of the two solu-

tion, which can be used to incorporate the "abnormal" evolu-

tions into a problem. The problem now becomes: when and to

tion brought about by the reversal of velocities, must take into

what extent do positions "count"?

account the fact that every collision creates correlations between


particles. It then becomes possible to say that the instantaneous
reversal of all velocities transforms the "post-collision" correlations, created by collisions during the normal evolution of the
system, into "pre-collision" correlations that, this time, determine which collisions will take place.

From this point of view, normal evolution, to which the


^ - t h e o r e m corresponds, and the abnormal evolution introduced by Loschmidt appear as extreme cases. One is associated
with the dilute nature of the gas, the other with the specific features of the initial state. In the first case, correlations "don't
count." A collision can, of course, be defined as "creating" post-

What, then, is the status of the collisions and correlations

collision correlations, but these have no effect on the frequency

that helped make Boltzmann's failure intelligible? I would sug-

of the collisions that follow. However, they will play a determin-

gest assigning them the status of a problem. The problem isn't

ing role if a reversal of velocities transforms them into pre-col-

objective in the sense that it would be dictated by the object,

lision correlations. We are then presented with the second case,

or subjective in the sense that it would designate the subject in

where collisions are determined by the correlations that char-

terms of what it does or does not know. It belongs to the cate-

acterize the initial state (resulting from the reversal of veloci-

gory of relations in the sense that relations precede the solution

ties). And in this case, the "abnormal" nature of these collisions

that distributes what belongs to subject and object, respectively.

is expressed by the fact that they are defined as "destructive" of

Loschmidt's reasoning belongs to the categoiy of solutions. It

pre-collision correlations. 6

assumes there are perfectly determined trajectories for particles and asks us to "see" collisions "undo" what other collisions
have done. It can then restate Boltzmann's velocity distributions
as an approximation, that is, it can assign temporal symmetiy
to the object and the ^ - t h e o r e m to the subject. Collisions and
correlations enable the kinetic problem to resist the dynamic
solution that denies any objective meaning to Boltzmann's
y?^-theorem. They help "slow down" Loschmidt's reasoning
process before it rushes toward a solution. A problematic space
can be constructed in which the collisions introduced by Boltzmann retain a privileged relationship with the evolution of the

The problem of correlations is obviously central for the


nonequilibrium statistical mechanics Prigogine proposed in
1962. For it is these correlations that allowed the space that
lies between the two extreme cases to be inhabited, the case
in which kinetic collisions suffice and the case in which all
dynamic properties are required. In an ionized plasma or dense
gas, we must take into account the interactions determined by
the relative positions of the "actors" in the space. Even during
so-called normal evolution toward equilibrium, kinetic reasoning cannot then be limited to collisions alone. Correlations as
well must also be taken into account. For the evolution leading

L32

BOLTZMANN'S

SUCCESSOR

BOLTZMANN'S

SUCCESSOR

'33

irreversibly toward equilibrium to be generally intelligible,

potential. The only role statistical mechanics plays in Gibb's def-

the collision must escape the categories of "normality" (which

inition is to make explicit, in a general way, the relation between

creates correlations) and "abnormality" (which destroys cor-

this thermodynamic definition and its probabilistic interpreta-

relations) and be capable of being described in general as both

tion, which affirms the omnipotence of dynamic theoiy. But it

creating and destroying correlations. It must be an actor in the

cannot be used, any more than thermodynamics can, to define

correlation flow. Th e general theoiy of nonequilibrium processes

anything other than the equilibrium state, which has become a

can then pose the general problem of determining how, in each

state of maximum probability.

case, pre-collision correlations are "destroyed" by the evolution

Gibbs starts with a Hamiltonian representation of the sys-

and what role post-collision correlations play in this evolution

tem, one in which positions and velocities play equivalent roles

before equilibrium is attained.

and where the concept of a collision, which problematizes that

The distinction between pre- and post-collision correla-

equivalence, therefore, plays no role at all. If we had an exact

tions appears here as an obligation. The "correlation flow" is, by

understanding of the system, we could describe it as a unique

definition, temporally directed. As such, it is not able to impose

being. More technically, it could be represented as a unique

the arrow of time on dynamics. Its function is to allow the gen-

point in "phase space," a space possessing as many dimensions

eral problem of irreversible evolution as such to be formulated.

as the system has degrees of freedom (the position and veloc-

And this problem is not organized by the dynamic conservation

ity of each particle). However, in the case of a large system of N

of cause in effect. The correlation flow is expressed in terms

particles, this exact representation is inaccessible. Therefore,

of creation and destruction, not in terms of conservation. The

the system will be represented by an ensemble: the ensemble of

problem of irreversibility as articulated by correlation flow is

all the dynamic systems compatible with the knowledge we have

focused on the question of "roles." What plays a role in such a

of the system under consideration. This ensemble will corre-

situation? What must be taken into account?

spond to a "cloud" of points in phase space. The evolution of the

Let's turn now to the second stable reference: the general


theoiy of equilibrium formulated by William Gibbs. This theoiy defined the "state" reached, in one way or another, by any
irreversible evolution that takes place in an isolated system or
one that is at equilibrium with its environment. Here, the formulation of the problem fully supports the abandonment of

system, if it were perfectly known, would be represented by one


and only one trajectoiy in phase space. It will be represented by
the evolution of this cloud, more accurately, by the evolution of
the "density," or "probability" function, p, which describes the
density of points in the cloud in each region of phase space.
Time evolution of the density defined by Gibbs is part of

Boltzmann's project, which was to provide a precise theoretical

dynamics. Even if we do not have an exact dynamic descrip-

meaning to processes associated with an increase in entropy.

tion of the system, the veiy fact that the density is defined over

This meant that models describing these processes at the

the phase space forces us to consider that each of the systems

molecular level should have the ability to overcome the limita-

in the ensemble is governed by the Hamiltonian equations

tion of thermodynamics, which does not concern evolutions but

and, therefore, has one and only one trajectoiy, which is per-

the definition of equilibrium states, corresponding to maxi-

fectly determined even if we are unable to calculate it explicitly.

mum entropy or to the extreme of some other thermodynamic

Gibbs's representation, therefore, can be considered something

134

BOLTZMANN'S

BOLTZMANN'S

SUCCESSOR

SUCCESSOR

135

of a solution. It presumes the existence of the trajectory, which

abandoning any reference to a "purely dynamic" evolution. It

means that the evolution of density over time must be conserva-

is sufficient to introduce a coarse-grained description, whose

tive: each point belongingto the ensemble remains one and only

technical definition would be that of a "mean local statistical

one point. The cloud of points behaves like an incompressible

description per block." This description does not try to define

fluid, which can change shape but whose "volume," the number
7

of points it includes, must remain constant in phase space. This


is known as the "Liouville theorem."

phase space in terms of points but partitions it into "blocks,"


or volumes of arbitrary but finite size. Regardless of granularity, each small volume corresponds to a given characterization

What does this mean for the description of equilibrium? The

being applied to an infinite number of distinct pointsdis-

essence, the goal, the touchstone of the equilibrium statistical

tinct from the dynamic point of view. And every time a point

mechanics is that the thermodynamic properties that describe

belonging to the ensemble representative of the system enters

a physical system at equilibrium must be able to be represented

a "block," all the points in that block are considered to belong to that

on the basis of the average value of dynamic properties. Sta-

ensemble. As a result, dynamic evolution is no longer governed

tistical mechanics must create a bridge between the dynamic

by the Liouville theorem. What should be described, from the

description of unobservable dynamic actors and observable

viewpoint of dynamics, as a dispersal of representative points

macroscopic properties. In the case of an isolated system, whose

throughout space, coarse-grained evolution describes as a con-

energy is defined as invariant of evolution, the density, pf(?, from

tinuous extension of the ensemble, a kind of block-by-block

which the average values can be calculated, corresponds to an

epidemic propagation. The behavior, p, is no longer that of an

ensemble known as the "microcanonical" ensemble, which

incompressible fluid but that of a fluid that expands continu-

comprises all possible states of the system characterized by the

ously until it reaches a maximum corresponding to p i;. Coarse

same energy (or Hamiltonian). In other words, the description

graining does the needed trick.

of equilibrium implies that all states compatible with the given

That irreversibility is obtained by the introduction of an

value of the amount of energy in the system have an equal prob-

approximation confirms the interpretation that assigns it

ability of representing that system.

merely probabilistic status. There is, however, one small prob-

But what of the evolution that leads toward equilibrium?

lem, which signals that a genuine coup has taken place: the tem-

How can the evolution over time of p result, at pC(J, in an exten-

porality of the expansion of the density function becomes purely

sion of the initial ensemble to an ensemble comprising all the

relative to the approximation, to the "granularity" chosen. This

states characterized by the same energy? There is a radical con-

is unimportant for the description of equilibrium, which corre-

trast between the Liouville theorem, which subjects density to

sponds to the end point of evolution and is thus independent of

a form of conservative evolution, and evolution toward equilib-

the time required to reach that end point. However, this negates

rium, which implies that, in one way or another, all the points

any possibility of making sense of "dissipative" properties,

compatible with the overall definition of the system (in terms of

such as cross section and relaxation time, which quantitatively

energy or temperature) acquire an equal chance of representing

characterize the temporality of the approach to equilibrium and

it.

have well-determined experimental values. As we have seen,

There exists a solution of great simplicity, which involves

it is precisely the agreement between the predictions of the

134

BOLTZMANN'S

SUCCESSOR

BOLTZMANN'S

SUCCESSOR

135

Boltzmann equation and the experimental value of such proper-

we come to the same conclusion: evolution is "conservative." A

ties for dilute gases that was used as an argument by Prigogine

classical point remains a point. A quantum eigenstate remains

to claim that this equation "should have" a profound meaning.

an eigenstate. No classical or quantum system, if governed by

The theory of Gibbs ensembles, however, doesn't predict or

the Liouville equation, can achieve equilibrium through the

preclude anything. It involves the physicist in determining

process of dissipative evolution.

carefullythe "correct granularity," the proper approximation,

How, then, can we express the problem of evolution toward

needed to "save" the experimental properties that characterize

equilibrium in phase space in such a way that the above impos-

the evolution toward equilibrium.

sibility doesn't have the final word? Is it possible to "slow down"

In 1963, Prigogine adopted a problematic space constructed

the logical transition that leads to this conclusion? It was the dis-

from two classic, and incompatible, "solutions" of the problem

covery made in 1956 (together with Robert Brout) of an answer

of the evolution toward equilibrium. One was taken from Boltz-

to this question that served as the starting point for Prigogine's

mann and made use of collisions; the other, from Gibbs, denied

work. Brout and Prigogine found that it is impossible to slow the

that collisions played any particular role at all and implied a

transition of the problem toward its conclusion as long as the

connection between evolution toward equilibrium and "coarse-

Liouville equation, or that of Hamilton or Schrodinger, explicitly

grained" evolution. He demanded from the first that it provide a

express the integrable character of the system they represent. For,

"profound physical meaning" for what the second defined sim-

in this case, they describe an evolution that can be decomposed

ply as extrinsic (choice of granularity).

in terms of independent modes and, what's more, characterized

Prigogine started from the time evolution of density, p. gov-

by their periodic behavior. Not only is evolution toward equilib-

erned by the so-called Liouville equation, which implies the

rium impossible by definition, the very concepts of collision and

ability to address both the classical and the quantum cases. At the

correlation, which imply coupling among the different degrees

time, statistical mechanics had been extended to quantum sys-

of freedom of the system, are excluded. It is possible, how-

tems, and the Liouville equation, derived from dynamics, could

ever, to formulate the problem of dynamic evolution by leaving

be formulated in quantum terms. There is nothing surprising

open the question of the integrable nature of the equations that

about this; it simply reflects the Hamiltonian heritage of quan-

determine it. This is how Poincare had formulated the problem

tum mechanics. In fact, the formalism adopted by Prigogine is a

of integrability: a given system was represented with reference

formalism of operators. Just as the time evolution of the quantum

to an integrable system, described in terms of "modes." Each

wave function corresponds to the application of the Hamilto-

of these modes corresponded to a degree of freedom of the sys-

nian operator, H, on this function, the time evolution of density,

tem and evolved independently, without interacting with other

p, whether classical or quantum, corresponds to the application

modes. As for the divergence between the reference system and

of the Liouville operator, L, on p. Except for some important dif-

the system described, it was represented by a coupling term

ferences, the Liouville equation can be used to describe both the

between these modes, which are, therefore, no longer indepen-

problem of the classical evolution of an ensemble of dynamic

dent. It is this representation that went onto be generalized. The

trajectories and that of the quantum evolution of an ensemble

problem of evolution toward equilibrium would be expressed by

10

of eigenstates of the Hamiltonian operator. And in both cases,

distinguishing a "free" term in the density function, where the

i38

BOLTZMANN'S

SUCCESSOR

components are deprived of correlations, and a coupling term

BOLTZMANN'S

SUCCESSOR

139

How can we now describe the flow of correlations that

that includes pre-collision and post-collision correlations. The

expresses the interdependence of po and pc? It is at this point

problem of "correlation flow" can then be expressed.

that supporters of dynamics turn away in disgust, because

In the formulation proposed in 1963, the Liouville operator,

Prigogine suggests a veritable formal alchemy, which, although

L, is not applied to density, p, but to the two components of that

it does indeed provide a formal expression of the Liouville equa-

density, po, which represents the "correlation vacuum," and pc,

tion, replaces this limpid reflection of the elegant simplicity of

which represents the "remainder," that is, the ensemble of correlations determined by the initial preparation or brought about
by dynamic evolution. Obviously, the two components evolve
interdependently, and it is the nature of this interdependence
that allows us to redefine the two "stable references" in terms
of which the problem of the general theoiy of nonequilibrium
processes might be expressed.
The equilibrium state assumes a very precise meaning here:
it is the state in which collisions and the correlations that collisions continuously create no longer have any effect on observable
properties and, therefore, on the density, which is now constant
over time. This means that, in this state, the component, pc,
which continues to evolve because the flow of correlations does
not stop, no longer plays a role in the observable outcome, no
longer contributes to the definition of observable properties.
The definition of density can then be reduced to the descrip-

dynamics with a succession of shifting reflections, decomposing and recomposing an image out of contributions whose identity will be redefined from moment to moment. Four new types
of operators are defined, which allow us to specify how the correlation flow contributes to the respective evolution of p0 and
pr. A characteristic of the evolution of p0 is that it maintains the
correlation vacuum. To it there corresponds a "collision" operator, which expresses vacuum-to-vacuum transitions (passing
through correlated states), and a "destruction" operator (the
destruction of initial, that is, pre-collisional, correlations). The
evolution of pr expresses the "creation" of correlations (from
the vacuum) and their "propagation." Correlations propagate
through collisions. If a collision occurs between two uncorrec t e d particles, a binary correlation is created. But a new collision with one of the two particles resulting from the first creates
tertiary correlations between the three particles, and so on.

tion of the "correlation vacuum" state. It is then a question of

The crux of the alchemy in question is, as I have indicated,

determining how the system evolves toward a state represented

to slow down the transition between the dynamic problem

only by this correlation vacuum, po. Here, the interdependence

and its solution, that is, to break down the problem by creat-

of the time evolution of po and pt. becomes central. Except for

ing "dwelling" spaces for the beings that are required by the

the particular case of equilibrium, the component pc influences

characterization of the collision event or, more specifically, of

the evolution of po. Before the system reaches equilibrium, this

the chronology of successive collision events leading to equi-

evolution is thus different from the evolution corresponding to

librium. This chronology helps clarify what is meant by the

a system without correlations. Thus, describing the approach to

approach to equilibrium, but it is not capable of imposing this

equilibrium becomes a question of describingthe disappearance

approach in opposition to the dynamic solution, which denies

of this influence. The formal distinction between the two com-

it in principle. Just as correlation flow implied the obligation to

ponents, po and pc, can be used to expose the implications and

distinguish between pre- and post-collision correlations, the

obligations associated with the evolution toward equilibrium.

definition of operators for the collision, creation, destruction,

134

BOLTZMANN'S

SUCCESSOR

BOLTZMANN'S

S U C C E S S O R 135

or propagation of correlations implies the obligation to deter-

ligible at any moment. Correlations, which collisions natu-

mine a direction for time. Anyone who rejects this obligation,

rally continue to create, are irreversibly "propagated" but do

which falls outside the field of dynamics, will consider Prigo-

not retroact, do not influence the frequency of collisions. The

gine's approach to be begging the question.

correlation vacuum at time t then depends solely on the cor-

To provide a more intuitive image of Prigogine's "chronolog-

relation vacuum at the initial moment. That is why, as we have

ical" representation, we could say that his goal is not to follow the

seen, the case of dilute gases helps distinguish two types of col-

evolution of positions and velocities that define this system but

lisions, those that belong to the "normal" evolution of a system,

to study the evolution over time of relations that define the prob-

where correlations create correlations that will have no observ-

lem presented by the evolution of the system. In other words,

able effect, and "reverse" collisions, belonging to the evolution

instead of being formulated once and for all, the problem of the

resulting from a state prepared by reversing velocities, which

identity of the system evolves overtime. Obviously, the relations

are defined as destroying correlations.

that define this identity must not be confused with interac-

There is another important result, which is that in the long

tions. Physical interactions such as gravity are neither created

time limit, it is possible, for any system, to write a separate evo-

nor destroyed, they exist. The only thing that changes overtime

lution for p0. For this limit, the evolution of p0 no longer depends

is the intensity of their effect. In contrast, correlations and the

on pr. In the long time limit, eveiy system reaches equilibrium.

relations they build up "normally" have no influence on the evo-

Therefore, Prigogine's 1963 proposal can be used to specify

lution of density, but in the case of "time-reversed" evolutions

what has "become insignificant" once equilibrium has been

their effect is dominant. In 196?, such a "chronological" repre-

reached. But, except for the case where the validity conditions

sentation was not seen as being irreducible to a dynamic solu-

of the Boltzmann equation (the molecular chaos hypothesis)

tion, to a solution that would annul the need to take correlations

are satisfied, it does not seek to define the equivalent of a n ^ -

into account, because it would treat evolution in terms of the

theorem, that is, a general definition of entropy valid through-

equivalence between cause and effect. But this representation

out the evolution of the system. Such a definition should resist

constructed a problematic space in which the physical meaning

the Loschmidt challenge: the defined entropy should increase

of the approximations used by physicists and the conditions of

even during the system's "abnormal" evolution resulting from

validity of those approximations could be identified. Approxi-

the reversal of velocity. In 1962, however, this result was not

mation no longer referred to a generalized "lack of knowledge."

(yet) being sought. The evolution operators allowed for a "gen-

It assumed an "intrinsic" sense, based on a differentiation that

eral theoiy" of nonequilibrium processes in the sense that they

intrinsically characterizes a system and allows certain terms to

could be used to specify the problems imposed by each of those

be considered insignificant in comparison to others.

processes. But they could not be used to extract a more general

In particular, the reasons for the validity, and the limits of


validity, of the Boltzmann equation are perfectly explicit. Its
domain of validity, the description of dilute gases, corresponds
to conditions for which the evolution of po can be considered
autonomous: the influence of the component pt. on po is neg-

lesson, to claim how, in one way or another, all those processes


helped increase the entropy of the universe.

BOLTZMANN'S

HEIH

143

subdynamics represents the evolution responsible for the final


establishment of equilibrium. 2 This particular subdynamics

Boltzmann's Heir

will then be able to serve as the "thermodynamic" represental ion of the system. The remaining subdynamics are referred to
as "nonthermodynamic," for what they describe may dominate
l be evolution of p for a given time but is fated to be "forgotten"
.it equilibrium. They can then be assimilated to "contingency,"
whose effects may be spectacular but are transitoiy. Dumbly,
obstinately, irresistibly, the system evolves toward equilibrium.
The separation into subdynamics proposed in 1969 corresponds to a principle of classification that is valid at every
moment, not only in the long time limit. The "macroscopic"
or "thermodynamic" definition of the system is no longer the

In 1963, Ilya Prigogine was Boltzmann's successor. Like him, he


accepted the verdict concerning the second law of thermodynamics and wasn't tiying to challenge the power relations that
imposed such a verdict. Statistical mechanics is subordinate
to dynamics (and quantum mechanics). It cannot challenge it.
After 1970, Prigogine became Boltzmann's heir in the sense that
he intended to resume the fight that Boltzmann had abandoned.
For the sake of the second law of thermodynamics, our conception of dynamics must be changed. It is no longer a question of
providing the necessary approximations with physical intelligibility but of demonstrating the need to "enlarge" dynamics. In
1963 it was a question of "slowing down" the problem in order
to distinguish its various components. Now it was a question of
challenging the veiy formulation of the problem.
This transformation began with a physical-mathematical
development that appeared in 1969, the transition from the use
of correlation flows to the study of the dynamics of correlations.'

result of an approximation, no matter how well established. It


now appears as a form of stratification generated by dynamics
itself. Both the definition of the macroscopic level and the dissipative evolution now claim to correspond to a dynamic (that is,
intrinsic) property: the increasing insensitivity of the evolution
of p to nonthermodynamic subdynamics. This makes possible a
formal definition of systems for which the distinction between
"macroscopic level" and microscopic description (dynamic or
quantum) independent of any approximation may be defined: a
necessaiy (and sufficient) condition is that the equation determining their evolution allows the transformation that leads to
the definition of the separate "thermodynamic" subdynamics.
An integrable system in the Poincare sense does not allow such
a transformationno classification into distinct subdynamics
is produced. The system is and remains fully sensitive to all the
details of all its states; it preserves the entire memory of its own
past.

What made this transition possible was a new physical-mathe-

Before further describing this new way of presenting the

matical result: it is possible to exactly decompose the evolution

problem, I'd like to comment on its implications and, especially,

of p given by the Liouville equation into subdynamics, each of

ask about the possibility of its being recognized as an "event,"

which was subject to its own law, in such a way that one of those

that is, as authorizing a new story. This story would relate that

142

144

BOLTZMANN'S

HEIR

Galilean physics had invented the first physical-mathematical


representation that incorporated time into the veiy definition
of its object: the instantaneous state of any "Galilean body" is
defined by the equivalence between the past that produced this

BOLTZMANN'S

HEIR

145

construction of new matters of fact to which a function would


refer, but the possibilities for redefining well-known "cases," it
is centered not on the "can do," on the power of description, but
on "how to describe."

state and the future it will produce. And the event associated with

Alluding to the Deleuzian definition of philosophical con-

the classification into subdynamics would reflect the possibil-

cepts to singularize the question of "how to describe" indi-

ity of a different way of incorporating time into the definition of

cates that this question, which relates to Deleuze's notion of

the object. It is in relation to the future, that is, the equilibrium


toward which the system evolves, that the evolving entities will
be defined. Correlatively, equivalence is no longer a pivot. The
"future" will, on the contrary, be defined by its relative insensitivity to certain aspects of the present.
It is important to bear in mind that the results formulated by
Prigogine and his team are formal in the sense that they demonstrate the possibility of an operation but provide no general
recipe, applicable in eveiy case, for completing the operation.
There are a small number of special cases, the "shared assets" of
specialists in statistical mechanics that, like Friedrich's model
in quantum mechanics, constitute an effective field of experimentation. These can be exactly described according to different
formal approaches and, therefore, allow those approaches to be
compared. But these cases "prove" nothing, they are the terrain
for what can be called "experimentation with functions," in a
sense that relates doubly to Gilles Deleuze. For, on the one hand,
Deleuze introduces a radical distinction between the concepts
created by philosophy and the functions that are the product of
science. However, he also defines the practice of philosophy as
"experimental," in that philosophy experiments with the way
the concepts it creates resolve or help resolve a problem. 3 The
concept of experimenting with functions, therefore, specifically
refers to the work of physical-mathematical invention, where
the question is asked of what a new type of physical-mathematical function can do, what it requires, and the obligations it
entails. To the extent this inventive activity does not involve the

"functions," is likely to resonate with the conceptual problems


of philosophy. But this singularity is correlated with another
concerning the processes of scientific innovation, namely, the
difference between what is, and what is not, recognized as a
scientific "event." Like philosophy, scientific innovation must
take place "among friends"friends of functions, not friends
of concepts. In this case, only those physicists who maintain a
"sympathetic" relationship to the problem of the irreversible
approach toward equilibrium can, at this stage, find the proposed solution of interest. And the only ones who maintain such
a relationship are those who are dissatisfied with the hierarchy
of physics, that is, the approximations that ensure the transition
from the "fundamental" level to the "dissipative macroscopic"
level. For those satisfied with this hierarchy, Prigogine's formalism "provides nothing new" because it doesn't enable them to
create new "matters of fact" presenting new properties, merely
to obtain known properties in a different way. And in fact, they
could also claim that all of Prigogine's formalism assumes a
redefinition of density, whose precondition is that time has a
fixed direction. But, from the viewpoint of dynamics, this is an
artifact, a hypothesis that has been added without justification.
Consequently, the stratification that reveals the dynamic significance of dissipative evolution is largely an act of prestidigitation because the equations of dynamics have not been deprived
of their power to deny any physical meaning to the difference
between past and future.
Yet, for there to be innovation in the scientific sense,

154.

BOLTZMANN'S

HEIR

BOLTZMANN'S

H E I R '55

(nearly) all "competent" scientists, and certainly not only

equations themselves. It is no longer a matter of slowing down

"friends," must be forced to recognize the superiority of the new

the process of solution so as to open a space in which it can be

proposal. Here we have a typical example of the way the history


of the sciences interweaves distinct temporalities. Time, long
and weighty, is here part of the definition of what is satisfactory

negotiated but of redefining the problem in such a way that it


leads "directly" to the sought-for solution, that is, a generalized
form of the ^-theorem.

and what is problematic, and this definition is always immedi-

This is the challenge of what Prigogine and his colleagues

ately connected with the ecology of practices, in this case, their

called the physical representation of dynamic actors. Dynamics

hierarchical organization. If the inability of dynamics to account

in the customaiy sense involves interacting entities, and those

for irreversible evolution had handicapped physics from an

interactions, identified by the instantaneous equivalence of

ecological point of view, the 1 9 6 9 - 7 0 hypothesis might have

cause and effect, are by definition indifferent to the direction of

become an event, might have constituted an innovation. In our

time. The veiy definition of "entities" and interactions will now

story it is a step, not the end of the stoiy.


Let us return now to the early seventies, when Prigogine took
a step that, in a different stoiy, might have been decisive. For it
was now an "enlargement of mechanics" that Prigogine tried to

be at stake as they themselves will be asked to determine how


the distinction between past and future is defined, thereby
affirming a world in which equilibrium belongs to the future.
The reader may recall that interacting entities in classical

interest his colleagues in, an enlargement such that the sym-

dynamics admit a privileged mode of representation, which can

metrical equations of classical and quantum evolution would

be actualized when the dynamic system is integrable: a cyclical

now correspond only to a particular representation. This repre-

representation in which interactions have been incorporated

sentation is, de facto, the only one usable in the particular case

in the definition of the entity and where the system can thus be

of integrable systems (or, in the quantum case, of systems that

represented in terms of the independent evolution of each of its

can be described in terms of eigenstates of the Hamiltonian).

degrees of freedom. The Schrodinger equation refers to the same

But in the "general case," described by the field of statistical

kind of ideal, but this time, it is the eigenstates of the Hamilto-

mechanics, where the concept of thermodynamic equilibrium

nian operator that correspond to the privileged representation.

has a meaning, this representation loses its exclusive character.

In both cases the ideal (or spectral) representation is defined

The idea of enlarging mechanics shows that Prigogine rec-

once and for all: entities defined as independent, not engaging

ognized that irreversibility implies questioning the most stable


assumptions of dynamics. As I attempted to show in connec-

in mutual interactions, preserve this definition throughout the


process of dynamic (or quantum) evolution.

tion with the definition of a "state" (see Cosmopolitics, Book II),

In contrast, the "entities" defined by the "physical" repre-

dynamics is a highly coherent edifice: accepting one element as

sentation (broken time symmetiy) have an evolutionary identity.

self-evident may obligate us to accept all of them. Correlatively,

The fixed point from which the representation is constructed

rejecting one of its consequences may result in questioning the

is not, like the = sign, given at each moment and conserved

entire edifice. It is here that the radicalization of Prigogine's

from moment to moment. It is located in the future. "Physi-

approach becomes most evidentthe shift from the question of

cal" entities are defined as evolving toward the identity they will

the solution of the equations of evolution to the question of the

have at equilibrium (if they are part of an isolated system). At

BOLTZMANN'S

148

HEIR

BOLTZMANN'S

H E I R 145

equilibrium, when the system is defined by average properties

representation, therefore, includes, at each moment, those cor-

only, these "physical entities" will mutually ignore one another.

relations associated with the "thermodynamic" subdynamic that

This mutual ignorance reflects both the result of an evolution

represents the component of evolution leading to equilibrium.

and the fixed point from which the evolutionary identity of the

All the correlations associated with other subdynamics, which

particles can be defined. Over the course of time, and irrevers-

will be "forgotten" at equilibrium but contribute to the evolution

ibly, the identity of the entitiestheir relative differences, what

toward equilibrium, especially the pre-collision correlations

they are sensitive tois continuously redefined, and it is this

associated with the initial preparation of the system, belong

redefinition that the equations of the dynamics of "broken sym-

to the complementary component ^p,,. These are not included

metry" describe.

in the definition of entities that evolve toward equilibrium but

This new statement of the problem is reflected in a new


P

definition of the vacuum of correlation, now written p0 (where


p

is the "physical representation"). In effect, the definition of

the vacuum of correlation at each instant constitutes the definition of the identity, at that instant, of the entities framed by the
physical representation.
The principle of transformation leading to the "physical
representation" can be expressed rather simply. 4 Unlike the
relationship of complete equivalence among all the "canonical"
representations that can be applied to a Hamiltonian system, the
representation resulting from this transformation is equivalent
to the dynamic representation only for what concerns macroscopic properties, those we can observe and which correspond
to an ensemble description of the system. This transformation
actualizes the existence of a form of "selection" that the physical representation will attribute to irreversible evolution itself.
It is based on the property of "separability" of the "thermodynamic representation of the system" established in 1969, which
formalized the possibility of a classification that distinguishes
between those correlations that contribute to the definition of
the state of equilibrium and all others. This separability has now
been incorporated into the definition of the new "vacuum of
P

correlations," p0, in such a way that its evolution describes the


evolution of the identity of "physical entities." 5
The definition of the vacuum of correlation in its physical

could be said to belong to the contingency of the path actually


followed in moving toward equilibrium. By definition, they stop
contributing to the description of the system when equilibrium
has been obtained.
In proposing this physical representation, Prigogine fulfills
Boltzmann's project. He is able to formulate an ^ - t h e o r e m that
is valid not only for the restricted field of dilute gases, but for all
domains in which the macroscopic level has an intrinsic definition. The entropy defined by the ^ - t h e o r e m for dilute gases
"benefited" from the fact that, in this case, the evolution of the
correlation vacuum in the customary representation already satisfies an independent evolution equation. The entropy defined
by the new ^ - t h e o r e m within the framework of the "physical
representation" expresses the generalization of this situation. It
has the properties required by any kinetic model of thermodynamic entropy: it reaches a maximum whenever it is a function
of the correlation vacuum alone, that is, when all the correlations associated primarily with the particular nature of the initial preparation have stopped contributing to the observable
evolution of the system. Thus defined, the increase of entropy
is no longer "merely probable," it is no longer contradicted by the
"abnormal" evolution resultingfrom the reversal of velocities. In fact,
the opposite is true. It makes explicit what the reversal of velocities dissimulated. In the customary dynamic representation, the
reversal of velocities appears unproblematic, a straightforward

154.

BOLTZMANN'S

HEIR

implementation of the time symmetiy of the equations for

BOLTZMANN'S

H E I R '55

The physical representation, therefore, can be used to

which pendular movement provides an immediate illustration.

express what Boltzmann had been unable to make explicit in

The intuitive analogy with a pendulum that appears to "return to

light of Loschmidt's objection: the difference between the "nor-

its past" while its velocity is spontaneously reversed is destroyed

mal" state and the state resulting from a reversal of velocities.

by physical representation. Whenever the pendulum "takes off

The probabilistic interpretation is unable to clarify this differ-

again" in the opposite direction, it is certainly the "same" pen-

ence. Relative to an instantaneous configuration, probability

dulum, governed by the "same" equations, whereas in the phys-

can be used to explain the privilege of the state of equilibrium: it

ical representation, the operation of reversal entails a radical


redefinition of the dynamic identity of the system and its constituent entities.
The physical representation proposed by Prigogine can be
used to illustrate what the reversal of velocities "does." It is not
the "same" system governed by the "same" equations of evolution, which seems to return to its past, it is an entirely different
system that continues its evolution. For the definition of a state in
this representation no longer refers to an instant but entails a
relation to a past that has already sorted correlations and to a
future equilibrium in which that sorting will be completed. Its
definition answers the question: how can the present contribute to the future? The reversal of velocities radically transforms,
although in transitory fashion, the answer to this question
because it restores to a past that is already "past" the power to
contribute to the future. The correlations produced by collisions, which had been integrated in the correlation vacuum,
that is, in the definition of physical entities, are transformed
into "pre-collision" correlations belonging to Ppc. This transformation is expressed as an entropy "leap," a discontinuous
variation in entropy, which corresponds to the redefinition of
both Pp0 and Ppc. If the system were approaching equilibrium,
the entropy would suddenly decrease, which does not contradict
the second law because it is not spontaneous but imposed from
outside the system. After the reversal, the entropy of the "new"

is the most probable state because the overwhelming majority of


the a priori possible dynamic states of the system correspond to
its (macroscopic) characterization. But probability is incapable
of expressing the difference between two kinds of nonequilibrium state: states whose evolution leads the system toward equilibrium and states whose evolution distances the system from
equilibrium. Both states are similarly "improbable," while from
the point of view of irreversibility they should be radically distinct. The situation is made even more unsatisfactoiy because
the probabilistic interpretation implies that for each state of the
first kind there corresponds a state of the second, which means
that both kinds of state have the same (im)probability. Indeed, for
each dynamic state resulting in the first kind of nonequilibrium
state, whose future is equilibrium, there corresponds, by the
reversal of velocities, a dynamic state of the second kind, whose
future leads away from equilibrium. Consequently, the probabilistic interpretation does not help us to understand why evolution toward equilibrium is privileged, nor why the first kind of
"normal" nonequilibrium state is easy to prepare, whereas the
definition of the second kind of nonequilibrium state implies
having recourse to a theoretical argument such as "what if the
velocities of all particles were simultaneously reversed?" It is
incapable because it is dependent on dynamics, which states
that all dynamic states are equivalent.
We can then say that, in 1973, Prigogine gave the problem

system again increases, whereas the "new" initial conditions

of irreversibility a scope that, had it been accepted, would have

determining this evolution are gradually "forgotten."

implied a "rewriting" of the histoiy of physics. This involved

BOLTZMANN'S

HEIR

BOLTZMANN'S

HEIR

I", I

that resulted in the classical and quantum Hamiltonian for-

symmetiy "emerge" from the description provided by classical


dynamics.

malisms in a way that distinguishes what the customary nar-

This second meaning is acceptable because emergence, here,

rative confuses: the two meanings of "first." Naturally, the

expresses the historical inscription of irreversibility or dissipa-

Galilean equivalence is "first," not only historically but also as a

tion cLsproblems for post-Galilean physicists. Words themselves

narrating Galilean equivalence and subsequent developments

de facto technical reference. In the new narrative, systems that

reflect this inscription: irreversibility denies reversibility and

can undergo dissipative evolution continue to be defined on

dissipation denies conservation. The "negation" follows from

the basis of their contrast with systems that empower Galilean

t he fact that the language of dynamics asserts symmetry and the

equivalence. It is on the basis of such systems that the language

conservation of cause in effect because it has been constructed

of physics, a s it poses the problem of irreversibility, is con-

on the basis of exceptionally simple situations in which Galilean

structed. More specifically, it is on that basis that the language

equivalence can be confirmed. From then on, the possibility of

defining irreversibility as a problem is constructed, because,

describing dissipative evolution must be expressed as a breaking

in phenomenological physics, chemistry, and in all kinetic

of symmetiy, and irreversibility seems to emerge "condition-

descriptions, irreversibility is assumed unproblematically. Bui

ally," if and only if the system satisfies certain conditions (the

Prigogine intended to deny this primacy its relationship with

possibility of constructing the physical representation depends

any claim to being "fundamental," from which eveiything else

on the possibility of defining a "macroscopic level," a possibility

would be derived by approximation. Galilean equivalence is not

determined by a precise "dissipative criterion").

the Ariadne's thread among dynamic phenomena that Leibniz

The construction of this "emergence" as a response to a

claimed. It does not have the benefit of designating the point

problem imposed by history has an immediate consequence: the

of view from which the labyrinth of these phenomena would be

breaking of the symmetiy of the equations of dynamics brought

exposed to secure exploration as guaranteed on the basis of first

about by the "physical representation" of particles must, by

principles. Or, if we wish to make use of a Platonic-Maxwellian

definition, result in two classes of symmetiy-broken equations,

reference, to designate the contrast between the cave in which

one corresponding to evolutions that situate equilibrium in the

appearances reign and the truth of the Queen of Heaven.

future, and the other to evolutions that situate it in the past. The

To the two distinct meanings we can assign to the "primacy"


of symmetrical description there correspond two distinct meanings of the term "emergence." Prigogine rejected the first: irreversibility does not "emerge" from reversibility as a property of
particular physical systems. On the contraiy, it is the motions
that allow us to claim the equivalence of past and futuremotion
along a smooth, inclined plane, a pendulum or an ideal spring,
the periodic movement of the planetsthat retroactively appear
to be particular and not representative. However, he accepts
the second meaning: the dynamic equations with "broken"

breaking of symmetiy must be accompanied by a statement that


denies any physical meaning to the second class of equations.
Objections are frequently raised about the need for such a statement: the exclusion of the second class is ad hoc, it cannot be justified; we have returned to the initial situation after a pointlessly
complicated detour, and so on. Boltzmann was forced to yield to
Loschmidt's objection: his "mechanical" model of irreversibility
depended only on the selection of particular initial states corresponding to the hypothesis of molecular chaos. Other initial
states should unquestionably cause evolutions corresponding

154.

BOLTZMANN'S

HEIR

BOLTZMANN'S

HEIR

'55

to "antithermodynamic" behavior, where the system spontane-

broken, it is always broken in two. It is the "initial" symmetiy

ously evolves away from equilibrium. In Prigogine's proposal,

that creates the necessity of selecting and, if we accept Prigog-

such selection takes place with respect to the equations, not the

ine's proposition, we could also say that, in so doing, this neces-

initial states. But isn't this proposition equally arbitrary?

sity transforms the meaning of dynamic symmetry in turn. For,

There is one significant difference, however. In dynamics,

this symmetiy does not express some ideal intelligibility but

the question of initial states and the question of the equations

betrays the dependence of ordinary dynamics on the particu-

of evolution do not have the same status. The initial state is, by

lar cases that have allowed it to escape the need to define how

definition, whatever the physicist or his ideal alter ego, Maxwell's

past and future differ. That is why Prigogine can claim that his

demon, defines it to be. The definition of this state is, therefore,

selective statement that equilibrium belongs to the future pre-

"free"it relates to the "preparing subject" or the contingency.

cedes any physical description, that it is not part of what physics

As for the equations of evolution, these are given, and govern

should demonstrate, explain, or justify, but what physics must

any initial state regardless. In this context, any limitation on

accept as a starting point.

the definition of states is automatically interpreted as affecting

If we were to bring out the philosophical "heavy artillery," we

the ability to describe or prepare, but incapable of restricting

might be tempted to refer Prigogine's selective claim to a tran-

the demon with 10 33 arms, capable of simultaneously revers-

scendental order, but its transcendental character, to the extent

ing the velocities of all the particles in a system. In contrast, the

that it refers to the breaking of symmetiy and the necessity of

selection imposed by the "physical representation" affects the

choice, is clearly relative to the history of physics. The difference

equations supposedly shared by the demon and the physicist.

between past and future finds its expression in dynamic terms

Therefore, the demon is not "free" with respect to the definition

in the selection of one class of equations and the exclusion of

that imposes this selection. More technically, the state result-

another. But the question this selection seems to imposewhy

ing from the reversal of velocities, the state whose definition

this class rather than anotherdoes not have to be asked. It con-

expresses the demon's sovereign freedom and omnipotence,

cerns a mathematical artifact determined by the definition of the

no longer challenges the selection that places equilibrium in

difference between past and future as the breaking of symmetiy.

the future. The meaning and the effects of this reversal, which

What the selection selects is something that physics literally does

express what the "subject" in the presence of the "object" can

not have the right to question because its experimental practice

do, are contained in the relationship to the future incorporated

assumes it. The selective statement that "equilibrium belongs to

in the equations of evolution with broken symmetiy. In other

the future," far from being arbitrary, would correct the arbitrary,

words, the selective claim "we live in a world where equilibrium

the contingent particularity of the histoiy of dynamics that took

belongs to the future" can be said to hold pride of place with

no account of the obligations associated with a coherent physi-

respect to any particular description, for no dynamic evolution,

cal description. It would, therefore, eliminate a possibility, that

no matter how demonic, will be able to contradict it.


No one would dream of contesting the fact that the formulation of two distinct classes of equations of evolution with broken
symmetiy is simply the mathematically inevitable consequence
of the symmetry of the initial equations. When symmetiy is

of an evolution that would contradict the second law, which is


nothing other than an artifact, the artificially maintained extension of a syntax that expressed the now unrepresentative singularity of the first objects of dynamics.
By 197?, Prigogine could claim that the requirements of

IG6

BOLTZMANN'S

HEIR

his program were satisfied. He had defined the general class

BOLTZMANN'S

HEIR

157

puts forth as complementary do not entrap physicists, do not

of dynamic systems for which the concept of thermodynamic

expose them to the temptation of defining a "matter of fact";

equilibrium has a meaning. He had shown that the second law

they remind them that matters of fact have meaning only with

does not require a probabilistic approximation for dynamic

regard to a well-defined possibility of observation. Here, the

description, but belongs to another type of dynamic descrip-

two "complementary" representations refer to two distinct ways

tion authorized by this class of systems. But did he define a new

of relating to a dynamic system, of formulating the problem of

kind of physical-mathematical "factish" capable of "repeating"

its behavior. But each of them authorizes the physicist to claim

the Galilean miracle? Was he capable of supporting the claim

that there is nothing arbitraiy about the way the dynamic system

that his equations with broken time symmetry owe nothing to

is represented: the system will satisfy whatever requirements

the physicist's standpoint? In fact, Prigogine adopted a posi-

its chosen representation defines. Thus, there exists a possible

tion he compared to Niels Bohr's complementarity principle.

history in which Prigogine's proposition would have been seen

The physicist can choose to accept the customary representa-

as enlarging and stabilizing that of Bohr.

tion given by dynamics, where the very concept of correlation is

However, that histoiy is not ours. In our history, the comple-

superfluous. This representation, the only relevant one for inte-

mentarity suggested by Prigogine has a weakness that is not the-

g r a t e systems, can be referred to for any dynamic system, even

oretical but historical: for the majority of physicists, his "either,

if it is practically useless when describing most of them. But if

or" cannot claim to possess the incontrovertible character of

physicists choose to continue to refer to it, they won't obtain an

Bohr's complementarity argument, which refers to a necessary

exact definition of equilibrium or any of the experimental prop-

choice among the quantum analogues of Hamiltonian variables.

erties that characterize the approach to equilibrium. If these

Why abandon symmetrical representation when it allows us to

physicists take an interest in equilibrium and the approach to

define equilibrium in terms of an approximation? Why not leave

equilibrium, they will have to explicitly incorporate the arrow

the approach to equilibrium to the field of "phenomenology," to

of time into the dynamic representation, which they can now

the construction of relations that are experimentally relevant?

do in an exact manner. The "factish" does not, then, define the

Why give it the power of redefining the scope and meaning of

way nature explains itself. It expresses the fact that this nature

the "attributes of the Queen of Heaven," of forcing her to assume

imposes a choice concerning the way it will be explained. This

the diminished status of being only one of two complementary

means that what we have referred to as the great achievement of

options? In accepting the transformation of representation sug-

dynamics, the coincidence between "description" and "reason,"

gested by Prigogine to replace approximation and phenomenol-

is not reproduced. More precisely, the way this coincidence is

ogy with exact definition, aren't we using a sledgehammer to

achieved no longer refers reason to nature "in itself." Reason is

swat a fly? 6

situated, it cannot be dissociated from the physicist's question.

One exception was Leon Rosenfeld. It was Rosenfeld's influ-

As we saw in Book IV, complementarity is not really popu-

ence that led Bohr to explicitly associate macroscopic mea-

lar among physicists. However, unlike Bohr's proposition, there

surement with the permanent, which is to say irreversible,

is nothing ironic about Prigogine's. The two representations he

production of a mark. As we saw in Book IV, Rosenfeld was also

154. B O L T Z M A N N ' S

BOLTZMANN'S

HEIR

interested in the work of Daneri, Loinger, and Prosperi, who


associated measurement with the evolution toward equilibrium. 7 During the last years of his life, Rosenfeld collaborated
with Prigogine and his team, whom he consideredand coming from Rosenfeld, the compliment carries some weightthe
8

author of a "non-trivial extension" of quantum formalism. For


Rosenfeld, Prigogine's result was deeply satisfying. It confirmed
Bohr's claims that the wave function and the Schrodinger equation have no physical significance independent of measurement
and that the definition of observables and macroscopic description were inseparable. But this result removed the suspicion of
arbitrariness created by Bohr's rhetoric. It is not the free choice
of the observer that defines what will be a measurement device
and what will be a quantum system. To function as a measurement device, a system must accommodate the definition of a
"macroscopic level," must be able to evolve toward an equilibrium that will be, since it is a measurement device, associated
with the production of a "mark." There is nothing arbitrary
about the observables that correspond to the definition of the
device; they are defined by the type of equilibrium to which
this "thermodynamic" system (that is, one that satisfies Prigo-

HEIR

'55

opportunities for relating theoiy to experiment, is largely determined by a form of aesthetic-historical evaluation. The idea
that a subordinate science like statistical mechanics, focused
on the study of systems with N degrees of freedom, about which
we obviously cannot expect an exact definition in terms of trajectories or wave functions, might provide the answer to one of
the "mysteries" of quantum mechanics is, for those uninterested in the question of the arrow of time, rather unpalatable.
That same problem is addressed by the grandiose hypothesis
of von Neumann and Wigner, for whom the observing consciousness determines the reduction of the wave function, and
the even more grandiose hypothesis of "multiple worlds" of
Everett and Wheeler, for whom every "reduction" corresponds
to the fact that the entire universe is multiplied into as many
universes as superposed states composing the wave function
before its reduction. These hypotheses are consistent with the
"revolutionary" aesthetic of twentieth-centuiy physics, whereas
the solution offered by Prigogine contradicted that aesthetic.
Doesn't his solution assume that we acknowledge a dimension
of the observable world that has been denied in "revolutionary"
fashion by the probabilistic interpretation of entropy?

gine's dissipative criterion) is susceptible. So, for Rosenfeld, the

Prigogine's quest, therefore, has continued. If it is to be

Copenhagen interpretation was freed of any hint of "idealism."

accepted by physicists, the arrow of time mustn't require that

Measurement is an intrinsic part of physics and not the deci-

they acknowledge that there is a "profound physical meaning"

sion of a human being. Its inevitable role in quantum mechan-

associated with the observables characterizing dissipative evo-

ics can from now on be expressed without any appeal to human

lution toward equilibrium. It must win them over in spite of

subjectivity. Quantum mechanics is unique in that it must rely

their refusal to recognize this meaning. To convince those who

on two complementary dynamic representations, one, explicit,

accept no obligations other than those imposed by dynamic

expressed in terms of the wave function, the other, assumed

description, it is dynamic description itself that must impose

implicitly by the reduction of the wave function, expressed in

a transformation of representation, one that will also confer

terms of broken time symmetiy evolution.


The various ways of resolving questions of interpretation do not interest physicists uniformly, and their interest,
except when it is constrained by an innovation that offers new

meaning on the arrow of time. What is needed is a factish that


repeats the Galilean miracle.

THE

OBLIGATIONS

OF

CHAOS

161

qualitatively different behaviors. According to Duhem, in the


case of such pathologies, the physicist must conclude that her

10

equations do a poor job of expressing the problem given that the


"mathematical deduction" they authorize is "forever unutiliz-

The Obligations of Chaos

able," regardless of the precision with which the practical data


can be determined in any conceivable future.'
Some seventy years after Duhem, Prigogine would reprise
the same type of argument, but in a different problematic
landscape. The pathological beings presented by Duhem have
become legion and there is no longer anything pathological
about them. They now form a well-defined class of systems, the
"chaotic" systems. Their "nonrobust" character in the face of
approximation is now referred to as "sensitivity to initial condi-

Every dynamic system is defined by a set of equations that,

tions," and this sensitivity can be characterized by a number, the

no matter how complicated the system, can assume the same

Lyapunov exponent, which must have a positive value.

"canonical" form. Whether or not we can explicitly write the

Strictly speaking, the contemporaiy notion of chaos is not at

solution of those equations, the trajectory of the system, we

all specific to the equations of dynamics. It is applicable to all

are required to say that every dynamic system is character-

deterministic equations of evolution, whether they are dissipa-

ized by one and only one trajectory. As we saw in Cosmopolitics,

tive or conservative. In the public's memory, it is also associated

Book III, Pierre Duhem had already fought against this notion,

with a computer simulation by Edward Norton Lorenz involving

which gives the concept of trajectory unconditional validity,

meteorological evolution, a form of dissipative evolution. The

independent of the contingent fact that we may (or may not) be

equations used by Lorenz were perfectly deterministic and cor-

able to write an explicit equation for that trajectory. For Duhem,

responded to a very simple meteorological model. But Lorenz

dynamic trajectory was a method of mathematical representa-

found that radically divergent evolutions resulted even when

tion, a construction that entailed the possibility of determining,

using initial conditions that were so similar that their differ-

based on the value of an ensemble of measurable properties, the

ence was not displayed by a computer. The "butterfly effect" is

evolution of the value of those properties over time. As such, it

the parabola of the practical consequences of the divergent evo-

will always have a mathematical meaning, although it might lose

lutions determined by those highly simplified equations. Long-

its physical meaning. In order to have a physical meaning, an

term meteorological forecasting is impossible; the growing

additional condition is required: the possibility of deduction

precision of measurement or the increasing power of computers

must be robust with respect to approximation. Duhem referred

will not indefinitely increase its scope over time because, in the

to the mathematical definition of "pathological" trajectories.

long-time limit, any difference, no matter how insignificant

For example, two neighboring trajectories, initially arbitrarily

a butterfly fluttering its wings thousands of miles awaycan

close, will always ultimately diverge, that is, will correspond to

"count." Of course, in itself the beating of a butterfly's wings is

160

i6?

THE

OBLIGATIONS

OF

CHAOS

not "responsible" for anything, it is not the "cause" of the divergence. Rather, it symbolizes, hyperbolically, the sensitivity of
the evolution to initial conditions, the way in which the evolu-

THE

OBLIGATIONS

OF

CHAOS

163

exponentially with time (a negative exponent means that the


corresponding distance will decrease, or contract, exponentially with time).

tion of two chaotic systems, initially arbitrarily similar, will pro-

The question of chaos is a question that cuts across scientific

duce behaviors that, over the long term, will become as different

practice both negatively and positively. Negatively, it condemns

as if their initial conditions had been randomly selected.

the hope of progress tied to the development of equations

As such, the butterfly effect can appear as a limit to forecast-

associated with increasingly detailed descriptions, for which

ing efforts, whether meteorological or those associated with the

increasingly powerful computers would compute increasingly

large-scale computer models often used in economics. It can

illuminating results. Positively, it introduces a new relevance for

also become a resource for renewed metaphors for games of

mathematics, one that no longer gives pride of place to physics.

chance and necessity, renewed in the sense that unpredictable

The invention of new questions, new distinctions, new criteria

behavior and rigorous adherence to causal explanations are now

would no longer be primarily tied to the question of processes

quite comfortable together. But for mathematicians it symbol-

that explain a behavior, because chaos marks the limits of the

izes a problem, not a result. More specifically, it introduces a

power such explanations provide. It would result in the creation

field of research that addresses the new questions required by

of a "nomadic aesthetic," wherein the relevance of mathemat-

chaotic behavior and the obligations they entail for the practice

ics would become local and circumstantial. That is why some

of simulation. That practice is now simultaneously crucial and


traitorouscrucial because the equations for chaotic systems
cannot be integrated exactly and must be simulated, traitorous
because, in the case of chaotic systems, no individual simulation
is representative.
How can the topology of the landscape of solutions be characterized? How can the transition from the fact of unpredictability
to the disclosure of a regularity underlying that unpredictability be made? Defining a chaotic system by a positive value of
at least one of its "Lyapunov exponents" is one of those new

authors see in it the promise of a mathematical invention of


the question of forms.* The wave that forms on the surface of an
ocean does not imply the theorization of the ocean as a whole but
organizes its own questions. The ocean thus loses its status as a
"cause," for which the wave would be the consequence. Rather,
it is defined solely in terms of the conditions that must be satisfied for the wave to be produced. And milieus other than the
ocean can provide meaning to those same conditions, in such a
way that the "wave theorist" will be able to travel wherever the
mathematical object "wave" can be actualized.

characterizations. The Lyapunov exponent doesn't characterize

Conversely, protagonists from the most diverse fields of

individual behavior but an ensemble of behaviors evolving from

science are inclined, for better and for worse, to grasp the new

distinct initial conditions, arbitrarily close to one another, or, to

mathematical questions and objects offered by chaos in order to

put it differently, it characterizes the topology of the space where

propose a transformation of the questions, modes of judgment,

those different behaviors will be represented as so many trajec-

and priorities in their field. And among those protagonists we

tories. Having a positive Lyapunov exponent means that in the

now find Ilya Prigogine, who challenged chaos to provide new

corresponding dimension of that space, the distance between

arguments in favor of the arrow of time. Chaos, if it is to be cap-

two initially neighboring trajectories will increase, or expand.

tured and redefined in the quest for a theory of irreversibility,

i6?

THE

OBLIGATIONS

OF

THE

CHAOS

cannot be dissipative chaos. Prigogine, who is best known for


his work on dissipative systems (operating far from equilibrium), was not interested in nomadic chaos, which has become
a common reference for far-from-equilibrium thermodynamic systems and many other models, from metabolism to the
economy. He would devote his efforts to the kind of chaos that
could be defined by equations that are symmetrical over time,
the kind of chaos, therefore, that may be able to provide reasons
for breaking the symmetry of such equations. And in so doing,
he would take an interest in a class of systems that had been the
subject of a body of work already considered classical in statistical mechanics, work done primarily by the Russian school of
mathematical physics led by Kolmogorov, Arnold, and Moser.
Although the work of the Russian school led to the defini-

OBLIGATIONS

OF C H A O S 164

From the point of view of Gibbsian statistical mechanics, the


uniqueness of these systems is that the points belonging to the
ensemble representing a given system, even if they are initially
concentrated in an arbitrarily small cell in phase space, will,
after a certain amount of time, be found "anywhere" (assuming
the conservation of energy, of course). This dispersion helps
explain why, in this case, the least approximation in tracking
the dynamic system (the finest "granularity") can transform the
Hamiltonian description into a description of evolution toward
equilibrium, where eveiy point in the macrocanonical ensemble has, when dispersion is complete, an equivalent probability
of representing the system.
Establishing a precise connection between the definition
of "collisional" systems capable of "physical representation"

tion of systems currently referred to as chaotic, this was not in

and the purely dynamic definition of, now chaotic, "K-flows"

fact their goal. They were focused on studying dynamic systems

was doubly interesting for Prigogine. It meant "escaping" a

that Poincare s theorem defined as "nonintegrable" (see Book

field defined by Boltzmann's kinetic approach, a field felt to

II). The Russian mathematicians were the first to see in Poin-

be unworthy by most physicists for justifying the challenge to

care's "resonances" something other than an obstacle blocking

"pure dynamics." And it meant entering the prestigious field of

the road that led from the problem of dynamics to its solution.

pure dynamics "by the right door," the one that might lead to the

They studied the problematic landscape of qualitatively differ-

abandonment of the identification between dynamic behavior

entiated dynamic behaviors brought about by those resonances

and the integrable systems model. Maybe this abandonment

according to their density in phase space. This meant that they

would enable him to establish a "purely dynamic" meaning for

were investigating a spectrum of complicated behaviors where,

the concepts of collision and correlation, concepts that were

based on an initial state, a (Hamiltonian) system could behave

merely "phenomenological" for specialists in dynamics given

regularly or randomly. From this point of view, the periodic

that such concepts lost their meaning whenever the integrable

behavior of integrable systems is nothing more than a par-

systems model was invoked. Couldn't the connection between

ticular case, defining one extreme of the spectrum, the case for

irreversibility and collisions, which in itself served, from the

which all behaviors are regular. At the other extreme we find

point of view of those specialists, as a motive for disqualifica-

the so-called K-flow Hamiltonian systems ( K f o r Kolmogorov).

tion, be retranslated in terms of obligations determined by

Regardless of the region, finite but arbitrarily small, of phase

"pure" dynamics?

space, that region will always include points that bring about

Additionally, if the general definition of dynamic behavior

"random" behavior. K-flows have the property of being "sensi-

and the integrable systems model could be radically disassoci-

tive to initial conditions" that will later be associated with chaos.

ated, the interpretation in "purely dynamic" terms of collisions

i6?

THE

OBLIGATIONS

OF

CHAOS

THE

OBLIGATIONS

OF

CHAOS

167

and the correlations they produce or destroy might cease to be

contract, whereas, in the other they expand (the Lyapunov expo-

"optional," coexisting with the possible that denies it. Irrevers-

nent is positive). "We could say that the baker's transformation

ibility would no longer hang on the choice of not using, without

is a nondynamic model of K-flows for, after a series of trans-

any theoretical justification, a mode of description that has the

formations, any ensemble of points initially concentrated in a

power to eliminate it. Prigogine's program, therefore, became

small region of the space will be found scattered throughout the

greatly radicalized. For now it was the construction of a post-

entire space.

Galilean factish that was in play. The approach to equilibrium

The baker's transformation has been frequently described.

and, especially, the collisions introduced by kinetic descrip-

I will limit myself to pointing out that it can be broken down

tions would have to become capable of supplying the dynamic

into two steps. First, the square space is "flattened," without

"reasons" for why they cannot be eliminated for the sake of rep-

changing its area, into a rectangle (distances along the horizon-

resentation in terms of cyclical variables, that is to say, in terms

tal coordinate are expanded by a factor of 3, while those along

of entities that evolve independently of one another. 3

the vertical coordinate are contracted by a factor of 2). The two

And yet the above summaiy is merely an outline. The ingre-

halves of the rectangle are then stacked on one another to re-

dients are there, but we haven't yet learned how to combine

create the square. As a result, depending on whether they were

them, to explore the obligations entailed by dynamic chaos if

in the left half or the right half of the initial square, the points

adequately presented. It was to a model created purposely to

experience two different outcomes. For points in the left half,

illustrate the expansion and contraction of distances char-

the value of their horizontal coordinate has doubled, while the

acterized by the Lyapunov exponents of chaotic systems that

value of their vertical coordinate has been reduced by half.

Prigogine and his team now turned. The baker's transformation,

For points in the right half, the value of the horizontal coordi-

transparent in that the distance between question and answer is

nate has doubled and decreased by 1, the length of the side of

minimal, would serve as a pure field of experimentation for for-

the square, while the value of the vertical coordinate has been

mulating problems and, in this case, as a field of confrontation

decreased by half and increased by V2, the length of the half-

between the concept of trajectoiy and that of evolution toward

side (or the height of the rectangle). In other words, depending

equilibrium.

on whether the expansion results in the coordinate of a point

The baker's transformation is what is referred to as a map.

having a value greater than that of the dimension of the square,

Unlike the equations of dynamics, it doesn't present the prob-

two divergent outcomes are possible. This divergence is illus-

lem of integration, for it is the entire "space" that undergoes the

trated by re-forming the square, but it is not dependent on the

transformation, and its effects can be monitored on every point

specificity of the model. It is inevitable for any transformation

in that space. A series of baker's transformations causes every

that introduces an expansion or contraction of coordinates in a

point in a square of surface one to undergo an "evolution" that

space that is conserved. And it is responsible for the eventual

has the two essential characteristics of a dynamic trajectory: that

fragmentation of any region of space: no matter how small, any

evolution is deterministic and reversible. However, this evolu-

region of space will eventually end up being cut in two, and those

tion is "chaotic" in the sense that, in one of the two directions

two fragments in turn, sooner or l a t e r . . .

defined by the square, the distances between neighboring points

As such, the baker's transformation allows for a simple and

i6?

THE

OBLIGATIONS

OF

CHAOS

rhetorically powerful continuation (in fact, a reinvention) of


Duhem's argument against trajectory. Regardless of the initial
precision with which the position of a "representative point" is
defined in baker's space, we can calculate the number of transformations after which this initial information becomes insignificant, after which, in spite of that information, we will have
to assign all the points in this space the same probability of
representing the system. After which, therefore, we will be in a
situation that defines equilibrium in statistical mechanics. The
increase in imprecision, however initially small, that gradually invades the definition of points, can almost be "visualized"
when the coordinates of the points are defined in binary terms
(a succession of os and is). Every transformation can then be
represented as a "shift": for each point, the first decimal of the
"expanding" (horizontal) coordinate becomes the first decimal of
the "contracting" (vertical) coordinate. The remaining ensemble
of expanding decimals shifts up one row while the contracting
4

decimals shift down one row. Because a point must be defined

THE OBLIGATIONS

OF

CHAOS

169

the first case, mathematical knowledge appeared to be an ideal


that it was possible to approach on a continuous basis. But chaos
institutes a qualitative difference between increasingly precise knowledge and the knowledge of "actually infinite" precision. For only the latter provides trajectory with a well-defined
meaning for chaotic systems. Chaos puts the physicist's back
against the wall. In order to preserve the reference to trajectory,
the physicist must refer to a knowledge that is inaccessible by
definition. Not that of a demon, whose knowledge may be arbitrarily precise but is still similar to our own. What is needed to
preserve trajectories is a knowledge that is independent of measurement, one classical metaphysics attributed to god alone. If
the physicist refuses to make the transition from the demon to
god, from a knowledge associated with practice, no matter how
perfect, to one that is associated with a reality "in itself," or that
is contemplated not from a godlike point of view but from god's
point of view, she has to admit that chaotic systems obligate her
to reject the concept of trajectory.

with finite precision, corresponding to a finite number of binary

However, this is merely an argument, not a physical-mathe-

decimalsn, for examplefrom the first transformation we can

matical construction. Here, evolution toward equilibrium is not

"see" the precision of the definition of the expanding coordinate

intrinsic, has no intrinsic temporality. The number of transfor-

decrease as it is reduced to n-i decimals: a decimal of unknown

mations after which the system is at equilibrium does, of course,

value has been "called in" by the shift. After n transformations,

incorporate the intrinsic dynamics of the system (the 2, \n?n +

the expanding coordinate is completely indeterminate. After

1). But it also introduces the number n, which characterizes the

the ?n+ith transformation, the representative point can be "any-

"granularity" of our knowledge. This is only the first step, the

where" in the square. Prigogine concluded that, here, the notion

sign that the baker's transformation is suitable for "experiment-

of a trajectory corresponds to an illegitimate idealization.

ing with functions" that allow us to put forth a "true" approach

The reason advanced by Prigogine in claiming that trajectory


cannot resist the chaos test was initially formulated as follows:
we have always known that the precision of our measurements

to equilibrium, that is, a perfectly intrinsic description of this


approach, one that doesn't make reference to the contingent
granularity of our knowledge.

is finite, unlike the mathematical possibility of defining "one

Here, I'll limit myself to presenting only those results

and only one point." This difference between physical repre-

(obtained in the early r98os) that best express the "bridge"

sentation and mathematics was of no importance for integrable

Prigogine now endeavored to construct between a physical rep-

dynamic systems, but it becomes crucial for chaotic systems. In

resentation based on the dynamics of correlations and "chaotic"

i6?

THE

OBLIGATIONS

OF

CHAOS

THE OBLIGATIONS

OF C H A O S

170

dynamics. This involved the construction of a baker's model of

between irreversible evolution and the baker's version of the

the second law of thermodynamics based on a redefinition of

Liouville theorem, which calls for the conservation of the mea-

both the "states" and the "equations of evolution" that together

surement of an ensemble, that is, the number of representative

break the "time symmetiy" of the baker's transformation.

points in that ensemble? How, then, can he "prevent" a series of

Naturally, the baker's transformation does not, strictly

reverse transformations from reconcentrating the points in an

speaking, define an equation of evolution. This explains my use

ensemble that a series of direct transformations has dispersed?

of scare quotes. However, it does supply an analog for dynamic

It is this sort of experimentation with functions that the baker's

evolution. Not only is it deterministica point resulting in a

transformation lends itself to and it is here that physical-math-

pointbut it is accompanied by a reverse transformation, where

ematical inventiveness can demonstrate its power, its ability to

the vertical coordinate expands and the horizontal coordinate

overcome contradictions. For the chaotic nature of the baker's

contracts. In the mathematical idealization of infinite precision,

transformation can be used to bring into existence mathemati-

where a state is represented by one and only one point, this

cal beings capable of being substituted for points, insofar as

reverse transformation allows for the equivalent of a return to

they are part of a representative ensemble, without the need

the initial state, analogous to the reversal of Loschmidt veloci-

to introduce approximation. Approximation always introduces

ties: points scattered throughout the space by a series of baker's

grain of finite size, regardless of its fineness. These beings,

transformations, if subjected to a series of reverse transforma-

however, function like points, they occupy no "surface" (tech-

tions, gradually come together to again form the initial cell. In

nically they have measure zero). But they nevertheless allow the

the representation with finite precision, reversal is of course

ensembles of which they are a part to escape the consequences

no longer possible, but the situation is still not satisfactoiy. A

of the Liouville theorem.

series of reverse transformations can lead the system toward

In the space defined by the baker's transformation, this

"equilibrium" as well as a series of direct transformations. As is

strange being, which now represents the state of a (chaotic) sys-

always the case whenever a coarse-grain procedure for the evo-

tem, is none other than a horizontal or vertical line or "fiber."

lution of density, p, in phase space is used, equilibrium can cer-

A line without thickness does not define a surface: therefore,

tainly be defined, but the connection between equilibrium and

like the point, it has measure zero. And yet, if a horizontal line

the "future" cannot.


This time the question is presented directly, with no escape

is subjected to a succession of baker's transformations, it will


be multiplied until, in the long time limit, it covers the entire

possible. Prigogine reached the point that led to Boltzmann's

baker's space. A vertical line will gradually be reduced to a point

failure, and needed to determine how the new ingredient found

when the number of transformations tends toward infinity. No

in chaos could change the situation. The baker's transformation

longer defined in terms of points but in terms of dilating and

seems to call for a description in terms of irreversible dispersion.

contracting fibers, the ensembles subjected to the baker's trans-

How can this be expressed without introducing approximations,

formation are able to confirm the irreducibility of their disper-

that is, without silencing one of the intrinsic properties of that

sive behavior. They can claim this behavior owes nothing to the

transformation, the fact that a point results in one and only one

human choice of "granularity": the "surface" (or measure) of a

point? In other words, how can he confront the contradiction

fiber being zero, it can, like the point, claim the ideal of infinite

172

i6? T H E

OBLIGATIONS

OF

CHAOS

precision. But this version of the ideal is now robust with respect
to approximation because the dilating fiber shares, along with
every finite region that satisfies the definition of finite precision, the dispersal process that is the baker's analogue of an
evolution toward equilibrium in the future.
Like physical representation ten years earlier, the description using libers entails a definition of the state expressed in
terms of broken time symmetry. As we have seen, although the
definition of the instantaneous dynamic state is indifferent to
the reversal of velocity, this reversal radically transforms the
state it affects if that state is defined by its "physical representation." Similarly, in the case of the baker's transformation, the
behavior of fibers radically changes, depending on whether they
are subjected to direct or reverse transformation. If subjected to
a reverse transformation, the horizontal dilating fibers contract
and the vertical contracting fibers dilate. Like the physical representation, the baker's transformation imposes a correlated
definition of transformations and states. And the baker's analog
of the second law of thermodynamics as a selection principle can
be precisely formulated, in the sense that it specifically excludes
the possibility of a particular type of state. Whenever we are
dealing with a series of direct transformations (which lead to
equilibrium in the future), states represented by the contracting fiber are physically excluded as they would correspond to
the initial state of a system that spontaneously and indefinitely
moves away from equilibrium. And the description based on
the concept of physical representation accurately reflects the
unique nature of this type of state. It is the only state where the
reference to equilibrium cannot be defined, which also means
that it is the only one whose physical representation cannot be
constructed. The kinetic analog would be the initial preparation of a system, described in terms of pre-collision correlations, that, rather than being transitory, "consumed" overtime,
permanently dominate the evolution of the system. 5 Using the

THE

OBLIGATIONS

OF C H A O S

173

second law of thermodynamics as a selection principle means


that some values of density, p, must be excluded, condemned as
corresponding to states that are physically unrealizable. In other
words, the various states defined by dynamics as being equivalent, indifferently realizable based on the free choice of the preparer, are no longer of equal value.
The legitimacy of excluding a categoiy determined by initial preparatory conditions can be defended on the basis of the
same principle that allowed Prigogine to characterize the trajectory of the baker's transformation as an illegitimate idealization.
The definition of such conditions cannot resist approximation.
Whereas evolution toward equilibrium brought about by a dilating fiber is robust, the evolution of a contracting fiber is not.
In the baker's case as in kinetics, the slightest lack of precision
is sufficient to reestablish the norm of an evolution that leads,
after an arbitrarily long period of time, to equilibrium. 6 And
the exclusion, moreover, is "propagated by dynamics": starting
from some admissible state, a system will never spontaneously
evolve toward an excluded state.
Prigogine's new argument, with its focus on the baker's
transformation, is simultaneously strong and weak. It is strong
because he can now engage directly with specialists in the theory
of dynamics. There is no further need to ask that irreversibility
be taken seriously, it must be recognized once a representation
can be constructed that is both robust with respect to approximation and intrinsic, that is, independent of the degree of
approximation. Expanding and contracting fibers satisfy these
requirements. They provide a sort of "intrinsic granularity"
independent of the choice of approximation. This means that
Prigogine no longer needs to depend on "sympathy," not for
himself, of course, but for his struggle for the second law. He can
argue a more general cause, in this case a rationality that is finally
relieved of its "irrational" reference to knowledge of infinite
precision, inaccessible in principle, and finally free to accept

i6?

THE

OBLIGATIONS

OF

CHAOS

the obligation without which mathematical representation has

THE

OBLIGATIONS

OF

CHAOS

91

In order for this argument to be taken seriously, whatever has

no physical meaning, a rationality finally capable of expressing

been realized through its use should apply to systems endowed

this obligation by inventing a reference that is stripped of all

with physical meaning. And in this case, Prigogine must con-

arbitrariness and is robust with respect to approximation.


It is weak for at least two reasons. First, the "limited rationality" that has become Prigogine's champion clashes, as did

front an additional difficulty: in quantum mechanics chaos, as


defined by "distances" whose time evolution is characterized by
the Lyapunov exponent, has no clear definition.

Duhem's claims before him, with the kind of realism that char-

The "realism" of the trajectoiy cannot be vanquished by

acterizes dynamics. As I have pointed out several times previ-

argument. It belongs to the register of those convictions that

ously, this realism is based on the fact that the dynamic object

can, whenever appropriate, hide behind a "yes, I know, but all

appears to dictate the way in which it must be described. Trajec-

the same." Nonetheless, this realism could be avoided, trans-

tory is then no longer a physical-mathematical construction, as

formed into an outdated conviction, if the interest of specialists

Prigogine implies when he argues for a different construction,

were engaged. But in order for this engagement to occur, it must

robust with respect to approximation. It is considered to be the

be fed with problems that the baker's transformation cannot

"truth" of the behavior of the system it describes. If, in the case

produce. We must return to Hamiltonian physics and Hilbert

of a chaotic system, we cannot access this truth, it nonetheless

space and subject them to the test of chaos. It is to this work that

exists because it is produced over time by the system itself. Refer-

Prigogine devoted the last years of his life.

ence to "divine" knowledge to which the physicist does not have


access is insufficient for disqualifying this realism. Such inaccessibility isn't a problem here because the physicist does not
conceive of her obligations in terms of knowledge. Because she
maintains a reference to trajectory, she feels obligated by the
very mode of existence of the system "itself." The system is, at
eveiy instant, "truly" in one and only one state, represented by
one and only one point. Any attempt to make her forget this "evidence" is merely a sophistic strategy designed so she will betray
her obligations. Against what, one could almost say is a heartfelt argument, Prigogine can certainly respond by introducing
other realist obligations: the description of broken symmetiy
is richer because it provides an intrinsic sense to the properties that characterize the evolution toward equilibrium, properties whose meaning we were, until then, able to construct only
by approximation. But the second weakness of his argument
now arises: the baker's transformation yields no observable
property because it is nothing more than a geometric model.

THE

LAWS

OF

CHAOS?

177

situations disappears. This is the case of "large systems," characterized by an infinite number of degrees of freedom in classi-

11

cal dynamics or by a continuous, rather than discrete, spectrum


in quantum mechanics.

The Laws of Chaos?

In quantum mechanics, continuous-spectrum systems are


physically very important but always calculated using approximation. They are important because they correspond to all situations where "something happens" in the quantum world, for
example, when we transition from an isolated atom, described
in terms of stationary states, to the excited atom interacting with
a field and characterized by a lifetime. The fact that approximation must be used to provide meaning to this "something happens" has, as readers may recall (see Book IV), aroused the

This time it was Poincare's theorem that defined the scope of the

criticism of Nancy Cartwright: doesn't the approximation that

problem, for it was this theorem that was most likely to contest

allows us to "leave" Hilbert space to describe the "causal" actor

the generalization of the model of integrable dynamic systems.

reflect the irrelevance of the static observables defined on the

And the obstacle it placed in front of that generalization was as

basis of that space? But the crucial point now is that, for the

valid for classical dynamics as it was for quantum mechanics.

ensemble of such situations, the Hilbertian guarantee according

As we saw in Book II, in classical dynamics Poincare's theorem

to which "there exists a spectral representation" disappears. In

embodies the impossibility, for the majority of dynamic sys-

fact, specialists of quantum mechanics are well aware that the

tems, of constructing a "cyclical" representation, synonymous

definition of Hilbert space ceases to be consistent once they are

with integrability, by "extending" the description of an integra-

dealing with a system with a continuous spectrum. More than

ble system through the calculation of perturbations. In quantum

forty years ago, the notion of a rigged Hilbert space was formulated

mechanics the situation is somewhat different. The beings that

to respond to this problem. 1 Here, "rigged" implies that Hilbert

inhabit Hilbert space not only satisfy causal requirements, the

space had to be "completed," "framed" by two additional spaces

reversible equality of cause in effect, their definition asserts the

capable of accommodating, as their legitimate inhabitants, the

quantum equivalent of classical integrability. The Hamiltonian

new kinds of actors introduced by continuous spectra. Hilbert

operator for a finite quantum system, one that is characterized

space is thus "rigged," the way one might rig a ship before it sails

by a discrete spectrum, can, by definition, be given a "spectral

across the broad ocean, far from the tranquil shores along which

representation," a representation of the Hamiltonian operator

it first traveled. But the status of these generalized spaces is well

in terms of eigenfunctions of that operator and its eigenvalues.

expressed in Leslie E. Ballentine's Quantum MechanicsIn

Poincare's theorem is an obstacle to the solution of a problem

beginning of the book (pp. 1 6 - 1 9 ) , the concept is presented and

(of integration) that does not seem to arise. Yet, as we shall

the author's summary concludes with the following: "These two

see, there exists one case where the difference between the two

examples suffice to show that rigged Hilbert space seems to be a

176

the

184,

THE

L A W S OF

CHAOS?

THE

LAWS

OF

CHAOS?

93

more natural mathematical setting for quantum mechanics than

"isolated" case, and it is nonrobust because the slightest pertur-

is Hilbert space." But in the pages that follow, this "natural set-

bation, the slightest coupling is sufficient to tip its description

ting" does not seem to entail any obligations at all: rigged spaces

into the general class of chaotic behaviors.

are only mentioned, and briefly, twice more in Ballentine's

The fact that in both the classical and the quantum case the

book. For it is approximation that now takes over and allows us

transition to the limit of large systems can be expressed as the

to "escape" Hilbert space when necessary.

loss of the reference model based on periodic behavior means

For Hilbert space to cease being the "natural setting" for

that "large systems" could indeed serve as the landscape where

large quantum systems is exactly what Prigogine needed. If

the meaning of Poincare's theorem changes, that is, becomes a

the obligations carried by the arrow of time or the ensemble of

crucial matter of obligation. It would not be simply an obstacle

properties with broken time symmetiy that characterize those

to the construction of a solution based on problematic equa-

large quantum systems are to be expressed in the form of new

tions, but would assume a positive meaning through which the

questions addressed to such systems, these must not, by defini-

appearance of observable properties with broken symmetiy

tion, be subject to Hilbertian requirements. A space offering the

would become "self-explanatory."

possibility of construction must be opened between the statement of the problem to which their behavior responds and the
definition of that behavior.

Prigogine would call all of these classical and quantum systems large Poincare systems, which makes explicit what he believes
to be their function. It is such systems that would become the

In Hamiltonian dynamics, the situation is equally interest-

reference countermodel capable of assigning its dynamic meaning

ing. What happens at the limits of large systems? In fact, the

to the arrow of time. It is such systems that would obligate phys-

entire economy of difference and similarity, assumed by the

icists to use something other than approximation when trying

representation of a "nonintegrable" system as a "perturbed"

to account for them. However, just as it is not sufficient to des-

integrable system, is overturned. Let us assume some strictly

ignate the "sensitivity to initial conditions" of chaotic systems

periodic behavior, that of an integrable system, for example: the

to overcome determinist reference, it is not sufficient to iden-

transition to the limit (number of degrees of freedom tending

tify the obstacle Poincare's theorem presents to the procedure

toward infinity) will have no effect on it. Now, take the case of a

of integration. Prigogine had to transform those "large Poincare

"nonintegrable" system, represented as a "perturbed" integrable

systems" into beings endowed with positive properties; he had

system in the sense that its degrees of freedom are coupled. In

to define them in terms of the requirements they had to satisfy

this case, regardless of the value of the coupling, no matter how

and the obligations they had to impose.

weak, the transition to the limit of the large system is expressed


by the appearance of chaotic behavior. In classical dynamics,
the KAM theoiy explored a diversified range of dynamic behaviors, limited at two extremes: integrable periodic behavior, on
the one hand, and chaotic behavior, on the other. For large systems, the spectrum is reduced to the two extremes, which no
longer have the same status. The integrable system is merely an

Two things are certain. To the extent that a bridge must be


made to the statistical mechanics of equilibrium, it is at the level
of an ensemble description that large Poincare systems should
allow a positive meaning to be assigned to Poincare's theorem.
And, in those cases that correspond to the approach to equilibrium, this should ensure that the requirements associated
with that approach are satisfied. In other words, an ensemble

THE

184,

LAWS

OF

CHAOS?

THE

L A W S OF C H A O S ?

94

representation must "tame" chaos, must turn it into an explana-

is being promoted as the key to a possible unification of physi-

tion for regular, entropy-increasing behavior.

cal interactions, requires calculations of unparalleled length

In physics, the "transition to the limit of large systems" is

and sophistication. A certain uneasiness has prevailed among

an operation that is both necessary and difficult. It is neces-

thousands of young physicists who had to confront a lifetime

sary because all the objects of statistical mechanics are "large

of scientific work devoted to such arid calculations. 4 But since

systems": a gas, a liquid, a solid only assume meaning at the

unification is the Holy Grail of contemporary physics, the price

"thermodynamic limit" (when the number N of particles and

had to be paid. With respect to the "transition to large systems,"

the volume Vof the system both tend to infinity while their ratio

however, this price is an obstacle, for the transition can appar-

remains constant). In quantum mechanics, as we saw earlier,

ently be carried out by means of approximations that are easily

the problems that introduce an interaction with a field (with a

controllable and reliable once they are guided by the phenom-

continuous spectrum) also correspond to large systems. But

enological properties we know we are looking for. That is why,

the operation is difficult in the sense that several properties

in Book IV, I emphasized that if the "Hilbertian factish" was a

must go to infinity at the same time (N and 7 for the thermo-

liar, its lie was interesting, for it was lending itself to the appro-

dynamic limit). When it is not a question, as it is in the statis-

priate approximations that served as a bridge between Hilber-

tical mechanics of equilibrium, of "building a bridge" between

tian properties and the phenomenological properties physicists

two known banks but of exploring the possibilities of a for-

need. The problem posed to Prigogine, if he was to interest his

malism, the limit transitions, if they are to be reliable, require

colleagues in the class of "large Poincare systems," was to make

extremely complex calculations, which, here, must be carried

the possibility of avoiding approximation and giving rigorous

out under suspicion: somewhere in the process of calcula-

physical-mathematical meaning to the transition to "large sys-

tion, haven't some pieces of "information " some possibilities,

tems" interesting in their eyesindependently of the question

been eliminated whose loss would explain the new properties

of the arrow of time, which for them was not an argument.

appearing "at the limit"? If this were the case, these would be no
more than artifacts of the procedure. 3 Of course, one can reply
that we "know" that new properties must arise: are the distinct
properties of the gas, liquid, or solid our artifacts? But there is
a risk that this argument may fail to convince, for the hierarchy
of physics has done its job. The critic is not interested in "common" phenomenological properties, she wishes only to remind
us, in expressing her suspicions, of the difference between a
rigorous "fundamental level" and an approximate phenomenological description.
The difficulty of an operation is not in itself an argument, for

At this point, something new intervenes: "large systems" can


be used to construct a method of description that can lay claim to
the beauty and intelligibility that give the Queen of Heaven and
her Hilbertian descendant their true power of seduction. The
new ingredient is a theorem by David Ruelle G986). According
to this theorem, the systems characterized by determinist chaos
can have several distinct spectral representations.5
How can such a theorem change the nature of the problem?
Recall that the customary spectral representation, a representation of the Hamiltonian in terms of its eigenfunctions and
eigenvalues, is the "ideal" representation once we make use

the evaluation of that difficulty depends on the value we assign

of a formalism involving operators. The evolution thus repre-

to what the operation targets. Thus, "superstring" theory, which

sented is as synonymous with beauty and intelligibility as cyclic

184, T H E

L A W S OF

THE

CHAOS?

L A W S OF C H A O S ? 95

representation in classical dynamics. In both cases, the rela-

It is with regard to the baker's transformation that Prigog-

tionship between problem and solution becomes limpid: the

ine will define the new object of his quest. The redefinition of

problem appears to dictate the mode of description appropriate

this object in terms of contracting and expanding fibers would

to its truth. The possibility of different spectral representations

be reformulated in terms of spectral representations: along

indicates the possibility of reinventing, for chaotic systems, this

with the "customary" spectral representation, correspond-

privileged relationship between problem and solution that is

ing to a description in terms of individual trajectories, there

characteristic of integrable systems. This time, the equations do

appear two new spectral representations that are probabilistic,

not define a system that is then found to have chaotic behavior,

that is, defined for ensembles. One represents regular evolution

they formulate the problem of behavior as chaotic.

toward an equilibrium located in the future, the other toward an

The definition of a spectral representation is the Grail of


mathematical physics; it obliterates the difference between the
register of the problem and that of its solution because the problematic definition of the object and the solution to the problem,
that is, the formulation of its law of evolution are expressed in the
veiy same language. We could just as easily change the metaphor
and say that, for Prigogine, the possibility of defining a spectral
representation with broken temporal symmetiy would shift the
grounds of the struggle from a war of insurrection, where the
rebel recruits allies who, in spite of their prowess as fighters,
are considered barbarians by those in power, to an attempt to
win power by means of a legitimate claim: the rebel finds he is
the heir of the power he is challenging, a direct descendant of

equilibrium located in the past. From this a decisive claim can


be constructed: the two representations with broken symmetry
cannot be reduced to the customary representation. It is impossible, even at the limit of perfect information, to "return" to an
individual formulation described in terms of trajectories. This
impossibility is crucial, for it implies that the new characterization is not simply another way of presenting the older characterization. And it is this impossibility that generally arouses
suspicion. If it is impossible to return to the "ideal" description,
it is because information must have been lost. Yet, in the new
case, the suspicion is excluded: the mathematics itself contravenes this reduction.
Here I want to give a very brief description of what otherwise

the great tradition of physical-mathematical invention. Like

might appear to be magic. Let's say that everything now revolves

all members of the royal line, the man who had previously "fol-

around the distinction between "regular" and "functional" func-

lowed" the phenomenological properties of matter, would have

tions, also known as "distributions" or singular functions. Such

succeeded in "getting the object to speak." Here we find Deleuze

creatures are now well known in mathematics, although they

and Guattari's distinction between itinerant and royal sciences:

remain fairly exotic in physics. Yet, it was a physicist, Dirac, who

"However refined or rigorous, approximate knowledge' is still

created the first of them, Diracs "delta," to help solve a quantum

dependent on sensitive and sensible evaluations that pose more

problem (which involved the continuous spectrum of a field).

problems than they solve: problematic is still its only mode.

Dirac's 8 has the strange property of being zero everywhere

What belongs, on the contrary, to royal science, its theorematic

except at one point where it is infinitely large. It was on the basis

or axiomatic power, is to isolate all operations from the con-

of the 8 function that the new mathematics of functional analy-

ditions of intuition, making them true intrinsic concepts, or

sis developed. And it is the fact that the 8 function must partici-

categories.'" 6

pate in the presentation of continuous spectrum systems, even

THE

184,

LAWS

OF

CHAOS?

though it does not inhabit Hilbert space, that is the origin of the
conception of rigged Hilbert spaces.
The question of irreducible spectral representations was
thus the result of the intervention of "functions" that originated in quantum mechanics but have since been endowed with
sophisticated new mathematical features in order to satisfy the
obligations of mathematical consistency. What they brought
with them was a dramatic restriction: an operator that implied
a singular function could not be applied to another singular
function.
This restriction has a spectacular consequence for the definition of what physicists conceive as the "ideal case," one that is
perfectly determined, where a description in terms of ensembles is reduced to the description of an individual system. The
density p becomes zero everywhere except at one point. This
means that from the viewpoint of the theory of ensembles, the
object associated with the "individual state" corresponds to a
Dirac 8 function. The point is no longer an intuitive object but
a singular limit case, and the primary target of the mathematical
restriction. And what this restriction affects are the new spectral representations that can be defined for chaotic systems, for
their evolution operator incorporates a singular function. Such
operators cannot, therefore, be applied to an individual system,
which also corresponds to a unique function. The description
of the "state" of a system can no longer be reduced to a point,
but only to an appropriate ensemble. 7 The ensemble description
no longer has as its ideal limit the description of an individual
system. The probabilities it introduces have the ability to assert
that they are irreducible and not relative to some contingent lack
of information.
For Prigogine, such a result was crucial, for it gave autonomy to the theory of ensembles, without which it is impossible
to construct a meaning for the approach to equilibrium. But
from the customaiy point of view of the physicist, it implies a
radical change of perspective. If the theory of ensembles were

THE

L A W S OF C H A O S ?

185

to become the site where obligations that have no meaning in


Hamiltonian dynamics or Schrodingerian quantum mechanics
could be formulated, it could no longer be defined as a simple
auxiliaiy formalism, intended to handle situations for which
we do not have perfect information. Correlatively, the nature
of a trajectoiy or a wave function would become problematic.
According to the conventional point of view, it is a question of
primaiy concepts, corresponding, in the case of trajectoiy, to
an intuition of motion. But in terms of the new perspective, the
primary concept is the ensemble. Trajectory and wave function become derivative concepts: trajectoiy can be redefined
as a unique object, localized, yes, but constructed in terms of
delocalized objects.8 The very term "delocalized" implies that the
ensemble has become the fundamental concept, for it is the
ensemble that can be said to be localized or delocalized. Location is no longer a primary attribute of members of the ensemble but a singular property of the ensemble as such, presenting
veiy specific types of problems.
For chaotic systems trajectoiy and wave function would now
be defined as the result of constructive interference between
delocalized plane waves, which means they would be defined
in terms of a representation, the object of which is an interference situation. And it is in terms of this interference situation
that Poincare's theorem can assume a positive value. "Poincare
resonances" no longer serve as an obstacle to integration, but
are used to explain the destruction of the interference that now
corresponds to the concept of location.
Prigogine showed that to construct a probabilistic spectral
representation with broken time symmetiy, the resonances
must correspond to "persistent interactions." 9 This last condition is trivially satisfied for all kinetic systemsit expresses the
need to speak of collisions in the plural rather than of a collision
as an isolated event. But the condition had to be made explicit
in the quantum case because some very important situations
did not satisfy it. The atom that spontaneously returns to its

184,

THE

L A W S OF

CHAOS?

THE

L A W S OF

CHAOS?

97

fundamental state can be represented outside of Hilbert space,

which the nonlocal definition incorporates, becomes a part of

and it is then described by a law of exponential "de-excitation,"

the problem and the point loses its status as a perfect definition

but in this case the spectral representation with broken time

of the state of a system. That is why Prigogine did not hesitate to

symmetiy brings wave functions onto the scenes, that is, repre-

claim, in the name of his representation, the most prestigious

sentations of the individual atom, not a probabilistic representa-

term in physics: this representation enabled him to formulate

tion.10 The de-excited atom implies a breaking of time symmetiy

the laws of chaos.

but gives no meaning to the evolution toward equilibrium.

Starting from the modest offer to take the experimental

Like the 197? "physical representation" whose results it con-

relevance of Boltzmann's ^ - t h e o r e m seriously, Prigogine now

firms, the "irreducibly probabilistic spectral representation"

presented the laws of chaos to his colleagues. Chaos, as the site

incorporates the breaking of time symmetiy in its equations of

of a new kind of spectral representation with broken time sym-

evolution (that is, in the definition of its evolution operator). 11

metry, would take its place within the illustrious history of phys-

As in 1972, this new definition of evolution brings with it a new

ical-mathematical invention, with the invention of beings who

definition of state. What changes, however, is the rationale for

have the unique ability to define how they should be addressed.

this new definition. It is no longer a question of taking the "sort-

For, when correctly addressed, description coincides exactly

ing" of correlations seriously, the fact that correlations become

with explanation. And the first heroic deed of this new "fac-

insignificant at equilibrium. The 197? definition of "physical

tish" would be to break the connection this invention had until

entities" was already nonlocal. Those entities were defined with

then respected, at least in the held of classical physics, between

reference to a future equilibrium where they could be described

"reality" as it is able to explain itself and "reality" in the "intui-

as uncorrelated (the specificity of initial conditions was forgot-

tive," "visualizable" sense that the Hamiltonian trajectoiy still

ten). But the nonlocal character of "states" to which the spectral

seemed able to identify. From the viewpoint of this intuitive

representation is addressed is now associated with the purely

reality, one may well wonder what is meant by the fact that the

dynamic and nonlocal concept of resonance. Consequently,

spectral representation of chaotic systems can be constructed

the nonlocal definition of state has the means to claim its rel-

only for ensembles. As a first approximation, the situation might

evance in purely dynamic terms, without having to make use of

appear somewhat similar to Brownian motion, the old symbol

an explicit reference to the state of equilibrium.

of the triumph of kinetic theory over the laws of thermodynam-

Correlatively, the abandonment of the ideal of the description of an individual system, the "localized" description, is
self-explanatory. It is the "local" definition of state, in terms of
a "point" in phase space, that becomes the impoverished definition, an artificially impoverished one, because it excludes any
reference to the topology of that space. This had no effect on
integrable systems, whose phase space is structurally trivial. But
when the "erratic" behavior characteristic of large Poincare systems dominates, reference to the topology of the phase space,

ics. The trajectoiy of a Brownian particle is "erratic," completely


irregular, but the corresponding probabilistic description is
perfectly regular: it characterizes the "diffusion" over time of
the probability of localizing the particle based on its initial position. However, unlike Brownian motion, the erratic behavior of
a "large Poincare system" does not appear to be the symptom of
underlying thermal activity. There is no "reason" for the loss of
relevance of the explanations that determine the behavior of the
individual system. A new type of explanation has been invented,

184,

THE

LAWS

OF

CHAOS?

together with a new type of being that dictates those explana-

THE

LAWS

OF

CHAOS?

98

context in which it retroactively acquires new meaning.

tions. In other words, if the Hamiltonian trajectory ceases to be

Retroactively, the structure of quantum formalism responds

a legitimate object, if the probabilistic behavior of the system

to a problem that has less to do with quantum beings than with

alone is the object of a law of evolution over time, it is because

what has been required of them. If an atom cannot be said to

what mathematical physics calls an "object," whether a trajec-

have emitted a photon except to the extent that a sensitive plate

tory or an ensemble, never had anything to do with a vision in

has recorded that photon, it is because "events," the evolu-

the first place, in the sense that vision is presented as being

tions during which "something happens," such as the emission

detached from any form of practice.

of a photon, have been used to construct the representation of

From the viewpoint of this new "realism," quantum mechan-

an object to which "nothing can happen," the stationary atom.

ics loses its status as a "privileged site" for philosophical reflec-

They have been "instrumentalized" as a means to access a phys-

tion, a site where the critical thinker might hope to find herself

ical-mathematical being that denies such a possibility. Within

on common ground with physics, for here "at last" physicists

the context of quantum formalism, the emitted photons do not

were forced to "think," forced to consider the limits of their

express the emission but the energy difference among station-

knowledge. Now, with Prigogine's hypothesis, the function

ary states. Once the specific character of this object construc-

could again "think" for the physicist. Which confirms Nancy

tion has been made explicit, which Nancy Cartwright has done,

Cartwright's argument about the problem of measurement:

the question of measurement loses its critical, epistemological

measurement, to the extent it had been reserved for a dialogue

dimension, capable of justifying the selective intervention of the

with philosophers, turned out to be powerless when confronted

philosopher. But as soon as this "something happens" acquires,

with the real problem, a practical problem corresponding to the

together with "large Poincare systems" and their spectral repre-

fact that physicists had to "cheat," had to introduce approxima-

sentation, the ability to define how it should be addressed, and

tions that enabled them to surreptitiously escape Hilbert space,

the ability to refer Hilbertian language to a specific case, one in

so they could "talk about" the experimental schemes in which a

which localized beings, described by a wave function, remain

given entity played a "role."

local, the question of measurement may appear to have been

Nonetheless, the laws of chaos have different implications


for classical and quantum physics. The abandonment of classical trajectory, its redefinition, in the case of chaotic systems,
as a singular object within the theoiy of ensembles, embodies,
as I have said, a reinvention that asserts the difference between
the physical-mathematical object and what intuition offers us as
immediately intelligible. In quantum mechanics, however, this
difference has, of course, already been asserted. It coincides
with quantum formalism itself. It is the very specific structure
of that formalism, with the duality between the wave function
and its reduction, which is now situated within a more general

resolved. Because it is necessary for "something to happen" in


order for something to be observed, the observation, which provides access to the quantum world, always implies a "large Poincare system." And this "something happens" must, as a result,
be characterized in terms of probabilities. Probabilities are not
then relative to a "reduction" that would, one way or another,
testify to the intrusion of the instrument into an unobservable
world, defined in terms of probability amplitudes. The instrument is not intrusive. Described as a "large Poincare system,"
it belongs to quantum reality. Retroactively, it was the choice of
defining observable properties satisfying the requirements of

184,

THE

L A W S OF

THE

CHAOS?

LAWS

OF C H A O S ?

99

Hilbert space, properties that referred to a reality where nothing

The question such spaces ask is: how, with what can they be

happens, that made probabilities seem relative and the mea-

rigged? From this point of view, Prigogine's spectral represen-

surement instrument responsible for whatever "happens."


It remains that probabilistic spectral representations do not

tations provide constraints and meaning. They show that what,


from the point of view of mathematics, imposes itself as a "natu-

dictate their reasons the same way that Hamiltonian or Hilber-

ral frame" can become an effective frame for the construction of

tian representations do. Maxwell's Queen of Heaven ruled over

an exact description of "large Poincare systems with persistent

a world that was supposed to "resemble" those rare cases where

interactions." This might be (or might have been) the "small dif-

his equations could be solved: eveiy system described in terms

ference" needed by a "great narrative" to accompany the "laws of

of dynamic equations "had" to have the same type of behavior,

chaos," enabling them to accede to the status of an event, and

the behavior exhibited by the periodicity of two-body motion.

creating its before and after. 12 The interest of mathematicians

And Hilbert space gave this claim the power of an axiom: all

would make a difference where Prigogine's "physical" reasons,

Hilbertian quantum systems can be represented in terms of the

rendered inaudible by the hierarchization of physics, were dis-

stationaiy model of the energy states for an isolated atom. How-

missed out of hand.

ever, the "new spectral representations" assert the constructive


nature of the solution they embody. For, the ability of a system
to define how it should be addressed and represented does not
preexist the construction of that representation. For each case,
"reasons" must be sought, its "intrinsic concept," as Deleuze and
Guattari would say. In other words, with the arrival of "singular"
functions, the general power of the = sign disappears. The factish proposed by Prigogine is deprived of the power to subject
dynamic systems to a form of "intuition" that recognizes "sameness" beyond diversity, an intuition that is, of course, physical-mathematical rather than perceptual, but just as capable as
perceptual intuition of masking the construction of the problem
with the evidence of a solution. That is why it is appropriate to
speak here of a new type of factish, the problematic factish.
Far from being a weakness, this feature of spectral representation (physicists refer to it as being model dependent) could
be of interest for certain key protagonists in the field of the
physical sciences, namely, mathematicians. For them, this feature may be situated within the general problem associated with
"rigged" Hilbert spaces, capable of accommodating the nonregular functions that arise with continuous-spectrum systems.

THE

PASSION

OF T H E

LAW

93

If the "laws of chaos" were ever to gain recognition, a decisive


ingredient would be the eventual alliance with actors who are

12

free in terms of the hierarchy of physics, mathematicians, who


were "amused to learn," as one said to me, that one of their exotic

The Passion of the Law

objects had acquired physical relevance.


Moreover, this is already the story told by the mathematical physicists who have agreed to look into "large Poincare systems": In the beginning was a mathematical being created to satisfy
the requirements of consistency that fall within the purview of mathematics. It vegetated in the respectable dignity of beings who were
irreproachable but without consequence, when one day it encountered the problematic terrain where its properties ceased to be exotic
and became necessaryfor the construction of a physical object. A new

Physical-mathematical invention is not pure. Like any inven-

chapter in the symbiotic relationship between physics and math-

tion, in order to be successful, to have access to the status of

ematics had been opened. A chapter where questions of kinetic

innovation, it must take into account the concrete landscape it

description, its role as a field of constructive exploration, will

wishes to modify. The story told here, independently of its out-

become part of the story, part of prehistory, while its domain

come, is significant in that the attempted innovation it relates

will be designated a privileged field of application, manifesting

is not presented as a "revolution" that overturns frames of

the fecundity of the laws of chaos.

knowledge and imposes questions that have gone unheard. The

If ever such a story were to be told, it would not be "false"

questions to which it gives meaning were not unheard of. It is

but it would feature an event that erases the passions and power

simply that they had been judged as unworthy of being heard, and

relationships involved in innovation, and becomes part of the

treated as such. For that reason, such attempted innovation also,

unbroken history of progress. It would give to mathematicians

and inseparably, has entailed creating and incorporating in the

and to their freedom with respect to the hierarchy of physicsa

object itself the means to appeal that judgment and, therefore,

role that would appear so "normal" that the potentially decisive

the power relationship that has historically allowed it to exist.

nature of their intervention would disappear behind the only

The arguments put forth by Prigogine in 1963 were associated

true subjects of the story, the properties of the physical-mathe-

solely with the experimental relevance of the Boltzmann equa-

matical being and the problematic terrain to which it responds.

tion: they didn't succeed. The theorization of collisions and cor-

I wanted to tell a different history, one that does not disqual-

relations suffered from the handicap of being associated with a

ify such "subjects," one that is not a contrary versionpsycho-

fieldkineticsdefined merely as phenomenological. The pro-

logical, social, or Machiavelliandenying the first. My history

posal to take into account the obligations of chaos and its sensi-

doesn't deny the "great story" because the eventual possibility

tivity to initial conditions encountered the split between what

of constructing such a story is one of the challenges imposed by

"we" are able to know and what a physical system in itself obeys.

history. It does not deny physical-mathematical beings their

19?

194.

THE

PASSION

OF T H E

THE

LAW

status as subject, for in the history I have related, they, and they
alone, could eventually gain the right to require and obligate. On
the other hand, it attempts to immerse this hypothetical great
story in the contingent and impassioned history that, in this
hypothesis, gave it the power to present itself as necessary.
In the eighteenth century, Diderot believed he could predict a
future in which the monstrous "pyramids" erected by mankind's
genius, that is, the work of the mathematicians, astronomers,
and physicists of his time, would instill not just admiration but
pity as well in those who would work to decipher, to conjecture,
to follow natural processes in all their diversity. This prophecy
is still current. It refers to a possibility that remains open. Concerning the history of mathematical physics itself, in Book III
I discussed how the history Prigogine continues to extend has
gone through a genuine process of percolation. The historical
reciprocal capture that coinvented, on the one hand, a physicist

PASSION

OF T H E

LAW

'95

the question, as such, is a determining ingredient of the ecology of practices? Does this mean, in other words, that I intend to
retain for physics the status it held, and continues to hold, whenever the expression "law of nature" is used? In other words, am
I claiming that the "laws of chaos" proposed by Prigogine are,
or may be, laws capable of engendering the diversity of forms
and regimes of activity that we observe, and as such of justifying a regime of peaceful coexistence among the various practices
satisfying that diversity? Wouldn't the "irreducible probabilistic representations" then simply be candidates for the position
of founder of a new physical authority, one that is more liberal,
more open but nonetheless the sole source of legitimacy? The
succession of these questions traces a predictable rhetorical
slope that 1 cannot ignore, for those first steps taken toward an
ecology of practices may serve as an enticement for following
such a slope.

capable of consigning, without scruple, to the field of phenom-

How can this slope be resisted when the narrative I have

enology anything that expressed the "arrow of time" and, on the

presented is haunted by a passion for law and the problem of a

other hand, the reality he addressedone capable of defining

method of expression that would confirm the authority of so-

how it should be addressed, that is, of dictating its own rea-

called fundamental laws over phenomenological description?

sonsmight never have taken place. But it did take place, just

How can we avoid celebrating a new hierarchization that is now

as the "capture" of the quantum world by Hilbertian formalism

simply more accommodating, no longer contrasting the ideal

took place. And it is on this succession of historical facts that

of perfect knowledge with the ignorance that qualifies all other

the antihistorical moral of the eventual "great story" I have pre-

kinds of knowledge, but still defining the "principles" of what

sented depends. If the reciprocal capture between "fundamen-

exists? How can we avoid presenting the eventual innovation

tal physics" and "a factish dictating reasons that deny the arrow

that woidd allow physics to no longer deny the arrow of time, to

of time" was to be challenged (which, for many physicists, is far

no longer reduce probabilities to a lack of knowledge, as good

from the case), the challenge would have to adhere to rules that

news, describing a physics that had finally produced the scien-

confirm the principle of the game, rules that demand that if a

tific truth of the possible, if not of becoming?

different kind of factish is to be defined, it must still be capable


of dictating its reasonsother reasons.
"What remains to be discussed is the importance I have given
to this kind of challenge as well as the question of the laws and
objects of so-called fundamental physics. Does this mean that

It should be pointed out that the possibility of falling into


this trap is included in the requirements Prigogine had to fulfill to achieve his ends. And I have wagered that a way to avoid
this trap was through the introduction of these requirements
as such. In fact, starling in 1973, by "following" Prigogine and

THE

194.

PASSION

OF T H E

THE

LAW

his group's research and trying to understand its implications, I


arrived at a constructivist conception of the physical object. And
I did so without at any time feeling that this conception would
diminish the interest of that object, quite the contrary. It is this
experience I have tried to share with my readers. My correlative
conviction is that physicists could, without in any way denying
the passion for truth that characterizes their practice, affirm not
1

the relativity of their truth but the truth of the relative. Although
the succession of theoretical factishes I have introduced, beginning with the first state functions expressing the power of the

PASSION

OF T H E

LAW

'95

liave as an ingredient the fact that the obstacle presented by the


"fundamental laws of physics" has been surmounted, that it no
longer functions as a sovereignty claim, demanding an act of
surrender from all other practices. However, the obstacle to be
overcome cannot be confused with the ecology being invented.
This should have as its actors all the practices it affects, each on
its own behalf, based on the requirements and obligations that
singularize it, and all of them, including physics, learning, with
the help of the others, to situate the truth of the relative they
inhabit and cause to exist.

Galilean = sign, are part of a histoiy that might have been differ-

Gilles Deleuze, quoting the poet Mandelstam, tells us that

ent, those factishes are not "soluble" in the contingent nature

"memory works not to reproduce the past but to set it aside." 2

of that histoiy. The requirement of a reality capable of dictat-

Which is exactly what I have tried to do. The past cannot be set

ing its own reasons might not satisfy any rational necessity; it is

aside by simple decree, it requires work, and this work cannot

nonetheless expressed by a set of specifications, or obligations,

be identified with a kind of critical deconstruction. It means

that physicists have found effective ways to satisfy. It is possible,

discovering and implementing the way it might allow itself to

therefore, because this would not require that they lose what is

be set aside without allowing it to be replaced with arrogant new

crucial to them, that they might agree to distinguish what they

certainties or a contemptuous indifference that substitutes for

have actually managed to construct from the meaning so often

thought. The challenge of my four explorations of physics has

given to that construction, namely, the disclosure of the "prin-

been to attempt to set aside a past in which the physics of so-

ciples" of reality.
This possibility, to which my personal experience has made
me receptive, has nothing to do with an argument that would
establish the plausibility or probability of a transformation of
this type, which would turn the physicist into a vector of peace in
the landscape of our practices. Moreover, to hope for this would
contradict my project given that such an argument would have
as its only subject the question of what physicists' practice could
make them capable of. This would be to confer upon physics,

called fundamental laws was identified with a claim as well as a


vocation, the discovery of "reality in itself," in opposition to any
"reality for us." But the "present" I have been working for confers no crucial importance upon physicists for giving up such an
identification. The fact that they can do this without having to
admit that their laws are "merely constructions" with no special
relation to reality is what matters for "us." For it enables us to set
aside a past in which the only choice appeared to be to accept the
authority of those laws or critically deconstruct them.

physics again and forever, and physicists' possible "goodwill," a

It is certainly tempting to critically address the fascination

key role, while my intent is specifically to bring into existence

exerted by the concept of "law." We could even speak of that fas-

relationships among practices that are purely immanent, which

cination, exerted on as well as by the physicist, in terms of the

cannot therefore be judged on the basis of any sort of tran-

passion of reason. But in this case, if this reason were to be iden-

scendence. The possibility of such an "ecology of practices" can

tified with "human reason," the past could not be dismissed. 3 To

194.

THE

PASSION

OF T H E

LAW

THE

PASSION

OF T H E

LAW

'95

avoid jumping to this conclusion, I had to show that we have no

our "vision of the world" and that Prigogine's hypothesis about

reason to think that the "reason" in question is human reason

a probabilistic determination of the present by the past, and the

in general. This is why 1 have related the very peculiar demands

future by the present, is interesting only to the extent that it is

associated with "reason" in physics to an event: the recipro-

not a general idea but a highly singular practical reinvention of

cal capture that turned the possibility of defining how the past

an object that satisfies the theoretical quest of physics. And its

determines the present and the present the future into a central

primary interest for "us," that is, from the point of view of an

feature of the physicist's vocation.

ecology of practices, is the realization that there is nothing gen-

From this point of view, there is a certain aesthetic satisfac-

eral about the very idea that it would be possible to define how

tion in the work of Prigogine and his coworkers. The definition

the past determines the present and the present the future, that

of "how" has not been directed back toward indeterminacy,

the attempt to implement such a possibility is what defines the

toward "free becoming." It has been transformed into a problem

singularity of the physicist's quest.

that must be solved for each individual case. In one sense, we

Leibniz, a mathematical genius, clearly saw the far-reaching

could say that qualitative diversity is back, no longer the diver-

consequences of the power to define the present as "conform-

sity of phenomena as in the nineteenth centuiy but the diversity

ing" with the past. 4 His strange definition of monadic reality is

of events and actors physics "brought into existence" ever since

one such consequence, and was confirmed by physics: changes

it decided to look beyond the diversity of those phenomena. Yet,

can be represented in terms of interactions and interacting

because it is compatible with a critical perspective, this kind of

entities, and also attributed to entities represented as autono-

argument cannot be used to "set aside the past." Because it takes

mous. 5 Prigogine's invention of a new type of "spectral repre-

advantage of the fact that a qualitatively differentiated reality

sentation" that introduces physical-mathematical beings again

resists the efforts of physics to tame it, there is a likelihood that

evolving autonomously, independent of one another, thus con-

such a perspective will initiate a new round of polemics, with

stitutes a new triumph of the affirmation of conformity, of the

physicists protesting that they do not tame reality.


My attempted narrative presentations concerning physical-

identifying power that brings together definition and reason,


description and explanation. But here the assertion of confor-

mathematical invention involve ways of defining the past that,

mity no longer has the power to conceal the fact that it answers

as far as the ecology of practices is concerned, must be set aside.

a question because the answer to this question does not possess

This does not need the consent of physicists. It involves, first

the "miraculous" power to erase the practical problem it started

and foremost, shifting the kind of interest generally associated

with. It has transformed the "case-by-case" identification of

with physics, an interest that connects physics with a mythi-

the approximations needed for physical relevance into a "case-

cal epic wherein "mankind" learns to recognize the nature and

by-case" construction of the object, thus depriving conformity

scope of the rights of knowledge, where it confronts the limits

of the power to tell the truth "hidden" behind the diversity of

and paradoxes of the confrontation between knowing subject

dynamic systems, all of which are defined by the same type of

and knowable world.

equation. Conformity is defined together with the effective con-

My attempt means claiming, without the slightest need to

struction of the solution to those equations. The object defined

downplay their interest, that the laws of chaos do not transform

by Prigogine satisfies categories that belong to the construction

194.

THE

PASSION

OF T H E

LAW

of the solution of the problem. If the resolution of a dynamic

THE

PASSION

OF T H E

LAW

'95

themselves positively with respect to physicists' demands that

problem continues to be identified with the construction of a

we recognize that physics defines the "laws of nature." To reply

"spectral" representation of evolution over time, with the royal

that these laws are laws of physics, not of nature, is not to reduce

road to constructing a factish capable of dictating its own reasons, the definition of the object can no longer silence its practical dimension, the requirements and obligations it satisfies.
It incorporates the question of how and at what price the past
determines the future in a way such that the answer can be provided only by the "case" the physicist is dealing with.
In other words, the laws of chaos have the interesting singularity of breaking the "intuitive" relationship that, even
when equations could not be integrated, allowed a well-defined
behavior to be assigned to the system, a behavior that would be
"self-deducing," "automatically calculating" its succession of
states from moment to moment. It is the construction of the solu-

them to arbitrary constructions. "Nature" is involved, yes, but


the way it is involved cannot be dissociated from what physicists are looking for. The laws of chaos are of ecological interest because they positively affirm what quantum mechanics
implied: the laws of chaos are not laws that "atoms" or any other
quantum or classical being would satisfy, the way Galilean bodies were said to satisfy the law of falling objects. And yet they
create a knowledge about reality as such, that is, about the extent
to which it isor is notcapable of satisfying the requirements
of conformity that Galilean objects fulfilled. And they do so precisely because they have accepted this requirement as the challenge they had to meet.

tion satisfying intrinsic mathematical obligations that deter-

What is an atom, a neutrino, or any other individual defined

mines the identity of the object described. As we have seen,

in terms of such a "fundamental law" Why can we get them to

satisfying the requirement of conformity affirmed by the laws of

"act" in our experimental devices, trap them, delegate them

chaos obligates the physicist to no longer ask about the behav-

as if they existed in themselves, whereas we can only define

ior of individuals but about the behavior of ensembles. And it is

them in terms of an ensemble representation? Let there be

these ensembles that allow the question of how the past deter-

no mistake, the question should not be construed as an objec-

mines the present to be answered. The laws of chaos say nothing

tion. Rather, it is at precisely this point, when the relationship

about what determines the reality they describe; they define the

between "fundamental," "objective" definition and the regime

extent to which, and on the basis of what definition, such a real-

of experimental existence becomes problematic, that physics can

ity is able to satisfy the requirement that brings description and

be said to create a knowledge about "reality" as such. Recall the

determination into correspondence. Yes, the quest continues,

remarkable agreement between the predictions resulting from

but this continuation inextricably designates the singularity of

the Boltzmann equation and the experimental measurements

the question and the invention that is required in order to sat-

that formed the basis of Prigogine's position: physics could

isfy the requirements associated with this question. The "laws

and should account for this agreement, should be worthy of the

of chaos" possess the truth of the relative.

experimental relevance of Boltzmann's model. But being worthy

The importance I have given to Prigogine's propositiona

does not mean explaining the relevance but inventing the means

present that would be able to set the past asideis thus an "eco-

to create consequences for the model, eventually transforming

logical" one. Whatever the verdict of his colleagues, his propo-

it into a first step toward an objective definition. Becoming wor-

sition entails the possibility that other practices may situate

thy of the experimental practices that have made atoms exist,

194.

THE

PASSION

OF T H E

LAW

act, become experimental actors and may mean that, but it is

THE

PASSION

OF T H E

L A W '95

(the metaphor concludes here: there can be no discussion or

only in terms of ensemble properties that we can define them

respect in phase space). In itself, the term "principle" expresses

"objectively," as objects capable of giving their own reasons.

the refusal to distinguish that singularizes this "resonant" phys-

Whatever the fate of the "laws of chaos," the problem-

ical heritage. Do we deal with principles of "reality," principles

atic character of the relationship between atoms, or any other

of any possible rational understanding, the principles of a voca-

experimental being, and the physical-mathematical factishes

tion that the physicist cannot betray without being deprived

that meet the requirements of objective definition is crucial

of the source of her invention? Bringing about the difference

from the point of view of an ecology of practices. Two distinct

between experimental factishes and the singular theoretical fac-

practices, with distinct obligations, are involved, making it

tishes that confirm all such principles, showing that experimen-

important to resist confusing "experimental factishes" with

tal existence and laws require distinct "capacities" from what

"theoretical factishes," especially those theoretical factishes

they address, means describing the respective achievements

that have, throughout the history of physics, extended the Gali-

those factishes constitute as situated in their respective spaces

lean miracle: the creation of an object capable of dictating the

of practice. The experimental neutrino, which has existed since

categories of its description.


In fact, it is hard to imagine a histoiy that is similar to our

the origin of the universe, does not obey the laws that claim to
provide the reasons for that universe.

own overall, but in which some analogue of Pasteur's microbes

Physics cannot, in itself, be the vector of peace in the sense

might never have seen the light of day, or an analogue of Perrin's

that the ecology of practices, as I understand it, gives to the term

atoms in the sense that they resulted in quantifying Avogadro's

"peace." A peace that is not tied to any form of surrender. A

number, or a molecule of DNA in the sense that bacteria can be

peace invented by each practice, along with the invention of a

made to bear witness to the particular role it plays. However

way to present itself that is compatible with the existence and

and this was the meaning of the metaphor of percolation I used

interest of other practices, satisfying other requirements and

in Book IIIthe histoiy in which the scope and significance of

obligations. But a step has been taken toward this possibility of

the "Galilean miracle" have been constructed, a histoiy that

peace, for the amalgamation of the neutrino of the laboratories

has included the Lagrangian event and its consequences, the

and the neutrino that appears to bear witness to the vocation of

probabilistic interpretation of thermodynamic entropy, and the

physics to decipher the principles of reality contradicted this

triumph of Hilbertian quantum mechanics, is characterized by

possibility. The theoretical-experimental neutrino, claiming

"improbable" points, which leave the imagination free to invent

the combined authority of its practical definition and its theo-

other possibilities. Possibilities in which Duhem, Bohr, even

retical definition, is incapable of coexisting with other beings

Engels, might have played rather different roles. Borrowing a

who create and are created by other practices. It can only pres-

metaphor from dynamics, I would say that at those points cer-

ent itself by referring those other beings to this "reality for us"

tain questions entered into immediate resonance, questions

that it transcends and judges. On the other hand, the fact that

that, while not strangers to one another, are generally discussed

the neutrino must be expressed in terms of two distinct prac-

in distinct terms, in the sense that the actors associated with

tical modes makes it possible to conceptualize the singular-

them generally affirm the need to respect their independence

ity of the questions and requirements to which, in each case, it

204

THE

PASSION

OF T H E

LAW

responds, rather than accepting the fact that it bears witness to


a "reality without us." The neutrino, by virtue of being expressed
twice, brings into existence two distinct and contrasting "us's,"

book vi

none of which may be identified with a general "us" that could


be contrasted to a "without us," because each of them is characterized by specific obligations. As such, these distinct "us's"
can be connected with the multiplicity of all the other "us's" that
are, simultaneously but each in a specific manner, producers of,
and produced by, what we call reality. The claims of the neutrino

Life and Artifice

belong to a space of coexistence in which all our creations allude


to the unknown in any creation.
This first step in the construction of an ecology of practices
can also serve to promote other questions raised by the generalized war of definitions that awaits us whenever we abandon
the physics of laws. Not in the form of a model, but to the extent
that we recall the way it came to pass. Not by criticism, disenchantment, or derision, which would confront a practice with
the need to abandon its hopes and doubts, dreams and fears, but
by acknowledging that the hopes and doubts, dreams and fears
associated with each practice make such an ecology interesting,
one that, although relative to histoiy, culture, or relationships of
social power, is capable of affirming the truth of the relative.

t h e

f a c e s

of

e m e r g e n c e

13

The Question of Emergence

If there is one question that transcends the field of so-called


modern practices, it is the question of life and artifice. Whether
technical artifice can produce life or only prepare the conditions
for it, while awaiting the breath that will animate material that
has been worked upon, whether the made being is faithful to its
maker or escapes its grasp, whether it escapes accidentally or
by vocation, or because the maker has "pierced" it, has partially
broken it to escape the monotony of manufactured products
these are the timeless stories, each of which reprises another,
older stoiy, that populate our memories. And within this fundamentally anonymous perspective, it is possible to situate the
impact of each new technique of delegation. From the medieval
clock to contemporary informatics and genetic engineering,
prosthetic devices, the synthesis of organic compounds from
inorganic molecules, or metabolic activities reproduced in the
test tube, every technical innovation capable of nibbling away at
the difference between our ability to have something do something for us and what living things do by themselves arouses the
same interest, the same confused passion, fear and pride. Every
time a delegated agent acquires a new skill, a new figure of the
living is made available for our stories, and a new figure of the

207

THE

2IO

QUESTION

OF

EMERGENCE

THE

QUESTION

OF E M E R G E N C E 108

risk assumed by those who dare to challenge the order of nature

their behalf that they signal a future in which, from psychol-

or creation.

ogy to the social sciences and therapeutic practices, all forms of

However, if there is one problem that is far from anonymous, that immediately brings up the question of the "science

knowledge concerning human behavior will be understood in


terms of neuronal interactions.

wars" with which the ecology of modern practices can today be

The question of emergence arises from this polemical con-

identified, it is indeed the problem of emergence. For in this

text. It was initially forged as a weapon against what would be

case, it is no longer a question of human power confronting the

called the reductionist bias. But any weapon can be used against

order of nature or creation, but the possibility, for a scientific

its inventor. The thesis of emergence sounded like a challenge:

discipline, of assuming power in a held previously occupied by

you cannot "explain" this emerging totality, as such, as the sum

some other discipline.

of the parts in terms of which it is being analyzed. Naturally,

Of course, we could claim that the question of emergence has

the challenge, once stated, was used to organize an explanatory

endured throughout the ages. Aristotle's disciples were already

counterstrategy. In other words, the theme of emergence trans-

arguing about composite bodies endowed with new qualities

forms the question of the obligations associated with the "emer-

that arose from the elements that composed them. How could

gent" into a held of confrontation. Will or won't this emergent

these new qualitative properties be explained? Was the form of

entail the obligation to "add" something to the operation of the

composing elements weakened or destroyed by composition, or

parts and, if so, does the addition in question entail the obliga-

did it remain untouched so that the properties of the compos-

tion to recognize the powerlessness of analytic thought?

ite would be novel in appearance only? We might be tempted to

In this context, the question of "laws," in the sense that we

claim that this same question was being asked in the eighteenth

speak of the laws of physics, is both very close and very far. It

century, when antimechanistic chemists claimed there was a dif-

is very close, in principle, because the reductionist argument is

ference between composition, which was their problem, and the

most often inscribed in a unitary vision of the world, in which

simple aggregation of physicists. 1 Except that, at this time, com-

the "parts" presented must, in one way or another, "obey the

posite bodies and aggregates had distinct spokespersons: the

same laws" as the matter studied by physicists and chemists. It

difference between them was now inseparable from the ques-

is very far, in practice, because no one dreams of requiring those

tion of the relationship between chemists and the supporters of

"parts" to actually bear witness to such obedience. 8 The hier-

mechanics. Similarly, when Leibniz pointed out the foolishness

archy that has already been established among the disciplines

of those who dreamed of explaining sensation, perception, and

here does its work. If chemical transformations and the ensem-

consciousness in terms of the mechanics of inert matter, he

ble of interactions among molecules are claimed to satisfy the

seems to have been taking part in a quarrel that continues today

fundamental laws of physics, any biological mechanism that can

with the unfortunately celebrated mind-body problem. Except

be analyzed in these terms should as well.

that now the quarrel is no longer a conceptual one; contemporary "materialist" philosophers no longer claim any status other
than that of being spokespersons for those who engage in what
would finally be a scientific approach to the brain. And it is on

On the other hand, reference to technical artifacts is far


more prevalent. As early as the seventeenth century, long before
the "science wars," the clock and the automaton, which had once
celebrated the splendor of divine creation, were used as part

2IO

THE

QUESTION

OF

EMERGENCE

THE

QUESTION

OF

EMERGENCE

211

of a philosophical operation that already forged the terms later

artifact whose making corresponds to another kind of practice.

used by scientists. The clock is a weapon against Aristotelian

At this stage, the example of the "defeat" of vitalist biology in the

thought, for which matter is unintelligible as such but requires

face of the so-called neo-Darwinist offensive, illustrated by the

a form, with which are associated both the existence of indi-

arguments of Jacques Monod and Richard Dawkins, provides an

vidual beings, each of which is endowed with its own end, and

opportunity to highlight the reasons why the ecology of prac-

the possibility of knowing them. In the case of the clock, mat-

tices I am trying to conceptualize must confront the question of

ter and finality can be understood separately: consisting of inert

emergence and, more specifically, the way in which this ques-

parts, and as such subject to the laws of mechanics, it owes its

tion has been defined in polemical terms.

clocklike existence to the genius of the maker, who has sub-

To what extent is the question of finality relevant to an

jected those parts to their own ends, who has incorporated them

understanding of living things? Only biologists not engaged in

into a coherent mechanism defined by a finalitytelling time.

the polemic with vitalism, such as Stephen J. Gould, were inter-

From Leibniz to Bergson, some philosophers were able to chal-

ested in asking the question, which then becomes very com-

lenge the relevance of the metaphor of the living organism the

plicated and very interesting, requiring fine distinctions and

clock proposes. But in the context of a "science war," it provides

risky hypotheses. Every characteristic can present a different

an inestimable advantage. The question of finality designates

problem, can tell a story that will distinctly interlink heritage

the stronghold that must be defended or conquered. For a cer-

and novelty, the coherence of previously stabilized meanings

tain time, the "teleology" inherent in the living has served as a

and unforeseen possibilities. We'll return to that. What I want to

standard for so-called vitalist biology. Although "mechanistic"

emphasize here is that understanding the challenge to which the

biologists might dissect the living at their leisure, organization

living being exposes the biologist is barred to the vitalist biolo-

toward an end would always be something that was "added" to

gist just as it is to the believer in neo-Darwinism. In both cases,

the dissected parts, which those parts cannot, as such, explain.

the polemical position is expressed by the production of an

Jacques Monod's well-known book Chance and Necessity: An

identity that is substituted for practical requirements and obli-

Essay on the Natural Philosophy of Modern Biology celebrated the

gations the way a solution is substituted for a problem. What the

fall of the stronghold. The teleological nature of living beings is

biologist deals with cannot pose the problem of the relevance of

only apparent, for they cannot be explained in terms of "final

the requirements in terms of which it is addressed, for, in doing

causes." However, they are "teleonomic," meaning that it is

so, the possibility of a betrayal, of a passage for the enemy, is

still on the basis of their finalityself reproductionthat they

liable to be created. As for obligations, these are mobilized by

allow themselves to be described. For it is natural selection

the supreme obligation of having the legitimacy of one's own

significantly referred to as the "blind watchmaker" by Richard

approach prevail.

Dawkinsthat must account for the singularity of living beings,


supply a reason for the characteristic ways a living being has of
reproducing, existing, behaving. 3

This mobilization, like any mobilization for war, introduces


slogans, watchwords. Thus, the case favored in reductionist
literature, the one that serves as proof and slogan, is the emer-

I will return later to the blind watchmaker, for he is currently

gence of the molecule of water, with qualitatively new proper-

being challenged by new protagonists, on behalf of a new type of

ties, from hydrogen and oxygen atoms. Similarly, the power of

THE

2IO

QUESTION

OF

EMERGENCE

THE QUESTION

OF

EMERGENCE

213

the laboratory should gradually dissipate the pseudo-problems

laboratory beings can in this way be mobilized for the polemic,

posed by the 'qualitative" emergences claimed by the adver-

that experimental questioning can be referred to as "reductive,"

sary. And this "holistic" adversary, partisan of the emergence

is one of the most damaging consequences of the science wars.

of a living "whole" irreducible to the sum of its parts, will, on

Where was the reductiveness when Pasteur had his microorgan-

the contrary, link his claims concerning the limits of experi-

ism "act" in a context such that its autonomy had to be recog-

mental practice to the fact that real "wholes" are proof that they

nized? Or when Korner, a student of Kekule, subjected the three

exist precisely to the extent that their properties can be objects

isomers of dibromobenzene to a substitution reaction, replacing

for description but not for experimentation. Thus, even if the

hydrogen with an N0 2 radical, the distinct isomers he obtained

scientist can intervene in the development of an embryo, she

demonstrating, by their relative proportions, the hexagonal

has to recognize the relative "autonomy" of that development.

structure of benzene? Or when the artificial DNA molecule syn-

Intervention can create monsters or kill, but the embryo can-

thesized by Nirenberg (UUUUUUU. . . ) succeeded, on May 27,

not be redefined in such a way that its development is proven

1961, in "causing"using all the necessary enzymes but "out of

to obey a function whose variables would be manipulated by the

the body " in a test tubethe synthesis of a protein, an obviously

experimenter.
Therefore, what the laboratory can do becomes the subject
of a polemic. For instance, it is the existence of a new type of
laboratory, that of the molecular biologist, that Jacques Monod
celebrated when he announced that the "secret" of the teleonomy of living beings had finally been pierced. The laboratory of
the molecular biologist has succeeded in turning living beings
into reliable witnesses, in subjecting them to the variables the
experimenter manipulates. Not living beings in general, however, bacteria and viruses. It is their performances that were
articulated in terms of the catalytic, regulatory, or epigenetic
functions of proteins, such functions relating to the associative,
stereospecific properties of those molecules, which is to say, in
the last analysis, to the DNA molecule containing the "genetic
information" that determines their synthesis. The partisan of

"stupid" protein, composed of a single type of amino acid? 4


Events of this kind mark the creation of new laboratory beings
and the new laboratories that correspond to them. 5 But they do
not pose the problem of emergence and do not allow any reduction to occur. They mark the success of an operation of delegation. The delegated being, which bears witness to its existence
(Pasteur), to what it acts on (Korner), or its specific responsibility (Nirenberg), brings about new practical possibilities.
Similarly, bacteria and the other laboratory beings that molecular biology has managed to turn into "reliable witnesses" were
in no way "reduced" to an arbitrary assemblage of molecules.
Those beings were targeted by operations of delegation or were
themselves delegated, and each of the "properties" that supposedly "explain" them celebrates the singularity of the successful
operation, not the generality of the power of explanation.

the irreducible emergence of the living organism (the "holist")

From this viewpoint, there is no need to try to determine

was betrayed by some of the living organisms he intended to

which characteristic would rightfully protect an elephant or a

represent. Now, he is asked to specify where, exactly, he claims

man from a reduction that would have succeeded with bacteria.

to break the chain of consequences that runs from the bacterium

It is much more interesting to point out how the operations of

to the elephant, not to mention humans.


That the invention of new kinds of laboratories and new

experimental delegation that have treated bacteria as targets or


actors have been made possible. The experimental invention of

2IO

THE

QUESTION

OF

EMERGENCE

THE

QUESTION

OF E M E R G E N C E 214

the bacterium takes full advantage of the fact that the bacterium,

emergence at all but are greatly interested in capturing success-

unlike the elephant or the man, undergoes no embiyonic devel-

ful operations of delegation and the claims of reduction that may

opment because it is "born" adult, whether in a test tube or any-

accompany them. This provides a twofold benefit: the power to

where else, whereas the elephant or the man need their mother's

create new ways of "doing" and the power to silence, in the name

womb. That is why the question of embiyological development

of "reduction" to an approach that is "finally scientific," those

is not the "same" question, only more complicated, as the mul-

who would contest the way the problem (for which these new

tiplication of bacteria. While bacteria have made it possible the

ways of doing supply a solution) is expressed.

impressive construction of the experimental factish of DNA,

The same situation occurred when the Pasteurian micro-

with its properties of replication, transcription, translation,

organism, vector of transmission of epidemic disease, became

and regulation, this factish does indeed possess the truth of the

the "cause" of that disease, the royal road to a "finally scien-

relative. It owes its autonomy to the experimental tests it suc-

tific" medicine that would reduce illness and healing to "purely

cessfully underwent, and this autonomy is therefore relative

biological" processes. This was a typical case of the reciprocal

to the tests the bacterium is able to experience from its envi-

capture of distinct interests. For doctors, reference to this royal

ronment without losing the stability of its definitionthat is,

road means adopting a position that gives them the power to

without dying. That the human embiyo or the elephant embiyo

disqualify charlatans. 6 For the majority of industries related to

cannot resist similar tests, that they require a "protected" envi-

medical practice, the difference between doctor and charlatan

ronment, does not protect them "by right" from future experi-

has little interest. But its consequence, the fact that the doctor

mental inventions. This difference signifies nothing more than

is made dependent on the network of laboratories providing her

that the question of how they are to be addressed will have to

with a guarantee of an "anticharlatan" practice, interests them

be invented. And if the precedent of the bacterium here had

much more. Medicine, like all modern practices, each mobilized

to serve as an argument, it would be to announce the possibil-

by conflict with other practices and all of them against opinion,

ity of surprises we are yet unaware of. For, prior to its experi-

is vulnerable to, and even demands, all the forms of capture that

mental invention, no one could have foreseen the extraordinary

ratify the validity of its position.

sophistication of the models it would impose, and continues to

In this joyous context, the fact that the "emergence" of mind

impose, on the biologist. For biologists, the question of deter-

in its relation to the "state of the central nervous system" may

mining "what a bacterium is capable of" is only just beginning.

appear to set the stage for a "summit" between science and phi-

When DNA becomes a "program," claiming to be the ulti-

losophy is a far ciy from expressing a privileged purity. Rather, it

mate explanation of all living beings and, at the same time,

is the glaring absence of any operation of delegation susceptible

claiming to give natural selection the role of a (blind) "program-

to reciprocal capture that ensures the disinterested character

mer," the sole (teleonomic) "reason" to which living organisms

of this "great question." The notion of state haunts the rhetoric

can respond, it is not the power of the laboratoiy that is being

of the sciences because it constitutes the master reference for

expressed but the power of the polemic that shaped the ques-

reductionist versions of emergence; but this reference indi-

tion of emergence on the field of confrontation. And along with

cates that emergence, in this case, is purely and simply defined

it the various powers that are not interested in the question of

in terms of confrontation. Confrontational reductionism has

216

THE

QUESTION

OF

THE

EMERGENCE

no need of the laboratory, and its relation with operational con-

QUESTION

OF

EMERGENCE

217

where there is no preexisting difference, won't it die of hunger

sequences and possibilities is simply a matter of rhetoric. The

in the midst of the two alternatives? asked those who wished to

only thing that really matters is that the adversary be disquali-

see it assigned the freedom, or will, associated with the ability to


decide "without reason." Leibniz had consigned the challenge to

fied, that he be lined up against the wall.


The "state" is in effect responsible for uniting "anything"

the ridicule it deserves. To satisfy the argument, the ass must be

that might be a relevant element of understanding the situation

represented as a pencil standing on its point. It is not at rest, but

and for expressing the possibility of organizing that multitude

"uneasy," any small difference will send it toward one alterna-

in such a way that "all" the relevant relations become relations

tive or the other. The "paradox" of Buridan's ass, which, absent

of reciprocal determination from which one should be able to

free will, will never choose one side rather than the other,

deduce a full description of what has become a "system." Ref-

implies not the fiction of two equally attractive pastures but that

erence to the state is typically followed by a challenge, with the

of a plane that would cut the ass, as well as the entire universe, in

adversary lined up against the wall. If he accepts that "eveiy-

two with no difference between the two halves. If the "mind" is

thing" has been accounted for in the definition of the state, will

to make "free" decisions, the "critical equilibrium" of neurons

he appeal in order to avoid the "reduction" to "something else,"

must also imply the entire universe. The universe, at this criti-

some ingredient whose sole meaning will be to express irreduc-

cal juncture, "waits" for John Eccles to choose between two pos-

ibility? And what is most curious is that this strategy "works."

sible futuresin one universe he will pull out his handkerchief

It succeeds in trapping some of those it targets. In The Self and

to wipe his nose and in the other he will sniffle.

Its Brain, John Eccles, wishing to "defend" mind, invents for it

Today we can anticipate a new quarrel involving systems

the ability to act through "infinitely weak" energy interactions

characterized in terms of determinist chaos. Doesn't the "state"

with large numbers of neurons "in critical equilibrium."" What

of a chaotic system lend itself to an even more convincing

splendid freedom it is to "choose" between two evolutions from

reductive argument? It fulfills all the conditions for reductiv-

some critical point. What an astonishing capability those large

ism because it is "determinist," that is, supports the claim that

numbers of neurons have that they are able to maintain them-

all relevant relationships are made available when determining

selves in "equilibrium" at some critical point in order to allow

the system's behavior. And because of the erratic character of

the "mind" the responsibility of choice.

this behavior, all of the manifestations an adversary might use

Eccles's speculation is representative of the astonishing

as indicative of the freedom to choose can be incorporated. This

intellectual regression provoked by the "science wars," a regres-

adversary will then have to show his true face (dualist, spiritu-

sion that explains why the mind-body problem is one for phi-

alist, irrational, believer . . .) because he will have to argue the

losophersand scientists who wish to "raise themselves up" by

difference between "true" freedom and behavior that is erratic,

addressing the "great questions." Eccles's presentation of the

unpredictable. The very terms of his argument will allow the

problem is none other than that already found in the old thought

reductionist to triumphally conclude that "we have left the

experiment of "Buridan's ass," which is faced with the necessity

domain of scientific rationality." Which means: we are entering

of choosing between one of two equally attractive alternatives.

the world of opinion, where anything is permitted but nothing

If Buridan's ass doesn't have the ability to create a difference

counts.

2i8

THE

QUESTION

OF

EMERGENCE

Hollow confrontations, power relationships, claims of constituting a royal road, complaints and accusations against the
conquering imperialism of a blind and calculating rationality,

14

visions of the world, and reasonall the confrontations that


serve as ecology in the modern sciences converge around the
question of emergence. Therefore, it is from this field of battle

The Practices of Emergence

that we must escape. More specifically, this field must be transformed into a problematic and practical terrain. But in order
to do this, the meaning of the term "claim" must first be transformed. Emergence cannot be disentangled from claims about
reducibility or irreducibility; therefore, a practical, constructivist sense must be given to the issues covered by that term.
It is not often that I have the opportunity to speak well of the
work of philosophers of science. That is why I don't want to miss
the chance to point out the parallel between the way I approach
the concept of emergence and its proposed definition by J. K.
Feibleman. He begins with a conventional definition of emergence, which associates the relation between a whole and its
parts to the relation between ends and means. According to
the "holist" version of this definition, the genuine "whole"
expresses its autonomy over the parts in that it can be seen as
its own end and its parts will be used as means to that end, or
purpose. Therefore, the "whole" is defined as being organized as
a function of that purpose. But to this conventional definition,
Feibleman adds an element that could change many things: "For
an organization at any given level, its mechanism lies at the level
below, and its purpose at the level above. This law states that
for the analysis of any organization three levels are required:
its own, the one below and the one above." 1 In other words, the
purpose of an organization is not found in itself but is always
seen from the point of view of something else.
As a test, let us apply this three-level definition to a favored
case of reductionism, the emergence of the molecule of water.
The interest in such a swing toward chemistry resides in the
319

222

THE

PRACTICES

OF

EMERGENCE

THE

PRACTICES

OF E M E R G E N C E 114

questions it brings to light, in this case, those that associate the

realization of coadaptation, the use of old properties that have

respective "identities" of whole and part to the practices that

been reinterpreted or new properties that have been ardently

allowed those identities to be defined. For, using the three-level

sought to show that the molecule can be fully explained by the

definition, the identity of water is immediately doubled, even

atoms of which it is composed.

within the practices of the chemists who defined it. Water plays

Another demonstration, veiy similar but this time involv-

two distinct roles: one of its identities corresponds to the chem-

ing statistical mechanics, would demonstrate the emergence

ist's purpose in understanding it as a molecule that will inter-

of the "whole" formed by a liquid consisting of a population of

act with other molecules; the other corresponds to the purpose

molecules. But the problem can be made more complicated.

of understanding it as a solvent, that is, a liquid. Consequently,

For the physical chemist is not the only one "for whom" water

"water" had to emerge twice: as a molecule composed of "parts"

is both molecule and liquid. The same is true for the living body.

and as a liquid with specific properties, composed of molecules. 2

Molecule and liquid "exist" for cellular metabolism in distinct

And, in fact, each of these emergences has three levels.

ways, each of which is defined by distinct purposes. In fact, the

Let's look at the molecule, while rememberingto distinguish

"purposes" of liquid water, as cellular metabolism as a whole

the atom from the chemical element. Ever since Mendeleev, the

constructs them, are far more subtle than those that made it a

element has been a part of the chemical definition of molecules

"solvent" long ago. Moreover, it is cellular metabolism that obli-

and reactions, but it presents no problem for emergence. The

gated the physical chemist to understand the subtlety of what

chemical element, like matter in the Aristotelian sense, has no

liquid water can do. 4 We can thus state the problem as follows:

properties that could be used to define it "in itself." Its defini-

from the point of view of cell metabolism, doesn't the "identity"

tion entails the definitions of simple and compound bodies and

of liquid water also "emerge" as being relative to the purposes

their reactions. The element does not explain the molecule, it

metabolism invents?

is explained along with it. On the other hand, the atom claims

The same type of problem can arise in the case of "detec-

to explain the molecule the way the part explains the whole. It

tion." It is not only from uncontrolled anthropomorphism that

owes its scientific existence to practices of a veiy different kind,

biologists talk about "detectors" when they describe a metabolic

which do not address it as a chemical actor; therefore it can,

function. In one way or another, living metabolism, as well as

unlike the element, claim a separable identity. "Emergence"

the laboratory, implies the construction of devices whose "pur-

can be reduced to two levels if and only if we adhere to Epinal's

pose" seems to correspond to detecting (assigning an identity

image of a chemistry that has been "reduced" to physics. In fact,

to) a molecule. 5 The irresistible character of the metaphor must

element and atom came to designate the same being only after

be taken seriously, but not literally. Perhaps, borrowing an idea

a series of complicated negotiations in which data from various

from Bruno Latour, who borrowed it from Michel Serres, we

practices had been articulated and coadapted. 3 And in this pro-

can make use of the prefix "quasi" to mark both the relatedness

cess of negotiation, the "purpose" is found "above," on the level

and the distinction between biological "practices" and practices

of the practice of negotiation itself. The identity of the molecule

of human understanding. A molecular quasi identity emerges

has been "organized" as a function of a known purposethe

from biological quasi detectorsa three-level quasi identity

THE

222

PRACTICES

OF

EMERGENCE

THE

PRACTICES

OF

EMERGENCE

223

given that it relates to the quasi purposes of detection and to the

confirmed by the appeal of its effects. The explicit recourse to an

quasi means constituted by the interactions among atoms used

ecology of practices that my definition of emergence expresses

by quasi detectors. 6
Let's return to the general question of emergence. If, as I

will have to shift the appeal that competing visions of the world
always promote.

have done, we include in Feibleman's definition the "purposes"

To assimilate, as I have just done, purposes associated with

associated with the practices of understanding, the question

practices of understanding with those that can be attributed to

assumes a practical and political sense. It signals a way of relat-

the living organism is somewhat forced, because the analogy is

ing two practices characterized by the fact that one includes

only partial. One way of making this partial character explicit is

in the definition of what it studies a reference to the object of

to point out the relative indifference of the experimenter (more

the other in the form of a "purpose," which is to say, it includes

specifically, the experimenters as a community) to the way in

the possibility of transforming what it studies into a means of

which "new water," redefined as a compound "emerging" from

explaining that object. In other words, the question of emer-

those parts known as hydrogen and oxygen, will redistribute the

gence is never "passively" asked, it is always actively asked. The

properties that could be attributed to old water. What is impor-

whole and its parts always refer to a third term, a practice whose

tant is the construction of a new story. The experimenters' appe-

purpose is to articulate their relation. Practice or quasi practice:

tite is now directed toward the creation of new devices, new kinds

the articulation of relations between neurons and the ways of

of proofs and tests, far more than on the means to "recover" all

experience do not interest neurophysiologists alone but had to

of water's former properties. The question as well of finding out

have been an issue throughout the history of living organisms

how composite water and solid-liquid-gaseous water are to be

with brains.

related is relegated to other research projects.

Once the question of emergence arises, whole and parts

The experimenters' appetite for the world from which they

must be mutually defined, negotiate among themselves what

take what will become the substance of their questions often

an explanation of one by the others implies. The holist version

assumes an aesthetic form. Thus, when Jean Perrin celebrates

of emergence denies the possibility of this negotiation because

the "vast host of new worlds" that atomic reality allows physi-

it identifies as a purpose for the "whole" the manifestation of

cists (us) to peer into, he also celebrates the defeat of values

properties that confirm that it cannot be reduced to parts. The

associated with "reality" by phenomenological physics, a reality

reductionist version of emergence transforms the negotiation

defined as regular, predictable, and measured by instruments

into unilateral determination because it is interested in the

that assume its homogeneity. 7 The thermodynamic phenomena

"whole" only to the extent that it promises to explain itself on

and their variables, which corresponded to laboratory prac-

the basis of its parts. It remains to be seen to what extent, when

tices that were quite distinct from those that, at the end of the

queried from the point of view of this negotiation, the ques-

nineteenth century, brought into existence the discrete world of

tion of emergence can cease being a battlefield where defini-

microscopic events "beyond phenomena," are, of course, said to

tions of "whole" and "part" confront one another, each claiming

emerge from that host. But it is this host itself that caused Perrin

both autonomy and the power to assign meaning to the other.

to speak, which made him a visionary and a poet. In fact, we can

This possibility, if it is to escape good intentions, can only be

go so far as to say that the question of emergence here is asked

222 T H E

PRACTICES

OF

EMERGENCE

THE

PRACTICES

OF

EMERGENCE

235

"backwards," for it is the parts that emerge from the "whole,"

lation can arise between molecular "means" and the "whole"

from the observable phenomenon. Contraiy to what has often

that constitutes the living organism. That is why, from the point

been claimed, there is nothing reductionist about Perrin, for

of view of selection, the purpose can be attributed indiffer-

whom discrete reality is not a "means" of explanation. On the

ently to genes or to the living organism. As Richard Dawkins has

contraiy, observable phenomena interest him only to the extent

stated, and his witticism is quite to the point here, we could also

that they are reinvented as a "means" for that discrete reality to

say that the organism is a means that the gene gave itself so as to

be characterized in observable terms.

ensure its own transmission from generation to generation.

The appetite of molecular biologists is quite different, but in

One of the most unexpected aspects of the "revolution"

this case as well the problem of emergence is presented asym-

known as molecular biology is to have created the concept of

metrically, privileging the means. When they subject bacteria to

"absolute" emergence, as Jacques Monod called it, satisfying no

tests that challenge their survival and their ability to proliferate,

reason other than that of selection. Like the clock, which owes

these biologists have effectively succeeded in occupying a posi-

to the laws of mechanics properties of secondary consequence

tion from which bacteria appear as being organized for survival

only, and everything to the intelligence of the watchmaker who

and reproduction, and the mechanisms they study then appear

made and assembled each piece, the living organism of molecu-

as so many means at the service of that purpose. But this posi-

lar biology is "compatible" with physical chemistry but owes

tion is unique. The role biologists have invented for themselves

nothing specific to it. Jacques Monod has never celebrated the

with regard to bacteria does not constitute a right for the scien-

prodigious activity of proteins and their interactions, but rather

tist with respect to the living organism. This role reproduces the

the cybernetic logic they obey. In fact, molecular biology, while it

one that bacteria are liable to confer upon the environment, the

celebrates the reduction of life to a gigantic network of catalytic

challenges they are capable of undergoing without necessarily

reactions, associations, and intermolecular regulatory activities,

dying from them.

also celebrates the triumph of technical artifice over Perrin's

The "vision" that confers upon DNAthe status of a program,


because it implies the omnipotence of selection in the role of the
blind programmer, assumes and affirms that the uniqueness of
bacteria is the truth for all living organisms. Eveiy living being
"says" the same thing as bacteria, except in a more complicated
way, and must therefore be able to confer upon its environment
the same kind of role. Regardless of the feature studied, its only
explanation is found in its selection: in one way or another, it
must have had a selective value, increasing its bearer's chances
of survival and reproduction. In other words, the power of
selection, which constitutes the "level above" from which the
living organism can be endowed with a purpose, can survive
and reproduce, would be limitless, so that no problem of articu-

teeming matter. It was not without reason that the specific performance to which proteins are susceptible has been compared
to microscopic "Maxwell's demons." Just as the demon embodied the rights of the probabilistic interpretation that enabled
it to intervene at the level of molecular activity and to impose a
form of collective behavior that broke with the rule of irreversibility, the performance of proteins subjugates chemical activity,
turning it into a biochemical "means" for achieving an "end" that
is foreign to it, that relates to a history of selection alone. Selection operates on a field that is always already defined by a logic
of subjugation since it operates on the result of unpredictable
mutations that primarily express the imperfection of subjugating chemical reactions governing the replication of DNA, the

222 T H E

PRACTICES

OF

EMERGENCE

THE

PRACTICES

OF

EMERGENCE

227

imperfection, therefore, of making those reactions subservient

as first formulated is only a hypothesis. Indeed, a "world" is

to the logic of conservation for which they are the means. 8

implied, and it will become an integral part of any "solution,"

Neither Jean Perrin, Jacques Monod, Richard Dawkins,

which may require that the problem be formulated differently.

or any other spokespersons of the all-powerful genes address

Of course, we can state that the experimenter also "envisages,"

emergence as a problem, thereby allowing the three "levels" cor-

but the space her practice delineates has a stable topology. She

responding to the problem to resonate. On the other hand, not

knows, a priori, what "the world" (that is, her colleagues) asks

far from them we can discern the figure of a true practitioner

of her, and she also knows what will identify a well-formulated

of the problem of emergence, whose appetite is stimulated by

problem. The obligations of proof, the creation of a reliable wit-

the possibility of emergence as such. This figure is the creator

ness, satisfying the requirements that put it to the test, supply

of technical artifacts, of beings who, if they manage to exist,

stable criteria for success. The technical innovator does not

will have overcome challenges that are associated not with the

know, a priori, how she is obligated nor what she may require.

requirements of competent colleagues but with the possibility

She inhabits a space for which a relevant topology must be

of reliable performance, endowed with meaning for an essen-

drawn, one that subsequently will be deciphered in terms of

tially heterogeneous collective and related to essentially dispa-

"means" that are implemented and "needs" that are satisfied.

rate constraints.

The question that orients the approach of the innovator (a

The technical-industrial innovator has nothing to prove, in

neutral term that refers to a group) does not fall within the per-

the sense that proof seeks to differentiate between fiction and

spective of discovery, and what is constructed has no ambition

fact. Her milieu is fiction. She is not, however, released from

to see any kind of preexistence recognized, the way a microor-

all obligations, quite the opposite. Her practice obligates her to

ganism or DNA might claim it. The delegated agents do not have

start with, if not create, a twofold indeterminacy. An indetermi-

to explain themselves, their actions do not have to bear witness

nacy regarding the way in which the being she creates satisfies

to the properties of corresponding actors. 9 They can do so, but

the constraints of the "level above," the level she addresses, that

that is not what is asked of them. Questions and agents respond

is, the level whose constraints that being will satisfy in giving

to one another within the perspective of a new emergence that

them a determinate meaning. And an indeterminacy regarding

must define both its prerequisites, what it requires of materials,

the way in which that being will distribute the respective values

of the processes and agents it is going to mobilize, and the way

of what it mobilizes from the "level below": what it will define as

in which it will be inscribed in the world, the purposes that will

a "means" and what it will define as a possible source of break-

identify it.

downs or problems to avoid.

Here the contrast between the possible and the virtual, the

The verb envisage is appropriate to this practice and its

real and the actual, found in Deleuze may again be relevant.

obligations. To "envisage" a problem does not imply its resolu-

As Deleuze noted, the virtual, has the "reality of a task to be

tion, at least not initially, but relating the terms in which it has

fulfilled." It is not just something that is susceptible to actual-

been expressed to the solutions it may authorize. The approach

ization, it confronts us with the problem of actualization. My

of someone who envisages is oriented, but not unilaterally.

earlier reference to the virtual concerned the question of quan-

It involves answering a question, a possible, but the problem

tum indeterminacy (see Cosmopolitics, Book IV). In that case,

222

THE

PRACTICES

OF

EMERGENCE

the "task to be fulfilled" was reduced to a mutually exclusive

THE

PRACTICES

OF

EMERGENCE

228

The scientist "in the field" is always on a specific terrain, never

choice between a determinate number of measurement possi-

one that can claim to represent all the others." The appetite of

bilities. With the question of the innovator, the virtual and its

scientists in the field in no way resembles that of the experi-

actualization rupture any relationship of nostalgia or mourning

mentalist, and those who study such scientists have to learn to

concerning the reality that would resist its "potentialization,"

develop an analogous appetite. For the stabilized operations that

the reduction of choice to a selection between already determi-

ensure judgment in the laboratory are also those that create the

nate possibilities. The innovator does not address a reality that

distance between the competent inhabitant of the laboratoiy

would be "potentially" defined by categories of knowledge yet to

and those who venture forth in this place where they know that

be constructed, to preexisting "potential" actors, lacking noth-

their questions will likely be judged idiotic, naive, and incom-

ing but the transition to scientific reality. Actualization is cov-

petent. The relative absence of such stability in the field can

ered by the "and . . . or" of distinct possibilities of emergence,

expose the one who studies fteldwork to temptations of ironic

rather than the "either . . . or" of mutually exclusive possibili-

relativism. Each scientist would define his or her "own" field,

ties of determination through measurement. Correlatively, the

all of them being equivalent before the ironic eye of the one who

irony of Copenhagen is no longer relevant. The " a n d . . . or" does

sees nothing other than the one thing that interests him, the

not impose abandoning a possibility. It brings about a new kind

power of fiction.13

of appetiteappetite for the "field" as speculatively implied by

The practice of the innovator spoke of emergence, not sci-

the possibility of emergence, a held where both the emergent's

ence. The practice of scientists in the field does not speak

requisites and the finalities that will be attributed to it must be

(directly) of emergence, for what is in play is first of all the

actualized.

question of how to "describe" rather than how to "interrelate."

Yet, while innovators are practitioners of emergence, their

As we shall see, however, the practical problems presented by

practice does not allow the question of what might be a practice

description have a direct connection with the question of a sci-

of emergence within the coordinates of science to be resolved.

entific practice that addresses the problem of emergence.

The technical-industrial-social factish to be constructed does

A geologist, a paleontologist, an ethologist does not "stroll"

not depend on the interest of colleagues, it has no ambition to

around, contemplating a scenic landscape; they do not explore a

raise new questions, to gather around itself practitioners who

place the way a photographer does, in expectation of an event, of

will connect it with other fields and other purposes. The appe-

the photograph that will be risked. They set themselves up with

tite for the field its construction brings about usually has a lim-

their equipment and their skill, and these specify their ques-

ited horizon as the success of the factish imposes the (relative)

tions and confer their meaning on the rather mundane photo-

stabilization of the purposes and means it distributes. 10

graphs they come back with.' 3 But what the field gives to them is

However, the appetite for the field characterizes sciences

not the answer to the question that such equipment and such skill

such as geology, evolutionary biology, climatology, meteorology,

refer to, but the description of a case, and nothing guarantees

and ecoethology, as sciences that address situations that cannot,

nor can guarantee that that case will serve as a reliable witness

as such, be "purified," reduced to laboratory conditions, that

capable of creating a trustworthy, and generalizable, relation

cannot, therefore, be reinvented in such a way that they become

between question and answer. Also, the answer is not capable of

(in some cases) capable of supporting a position of judgment.

being subsequently stabilized and narrated economically, as is

THE

222

PRACTICES

OF

the case after a successful operation of delegation, or any other


experiment. The answer provided cannot economize reference
to this exploration, carried out on this field. Nonetheless, we can
speak of an answer provided by the field because of the learning
such an answer entails, learning that does not result in conclusions but in narration.
Unlike the experimental factish, which, by definition,
"explains itself" in the answer to the questions it authorizes, the
field induces and nurtures questions, but it does not supply the
ability to explain the answer that will be given to them. Of course,
the practice that causes it to exist and is addressed to it assumes
that the relationships that allow themselves to be deciphered
are "conditions" for the answer, but they are insufficient conditions. However, the loss of the determining power of the condition, the fact that it is incapable of providing an explanation, are
not negative categories here. For there to be a field, the indeterminacy must be interesting as such, the questions addressed to
the field and the relationships it articulates must welcome the
possibility of a mutation of their supposed meaning. The needed
appetite for such a possibility and the role played by the field,
which is liable to lend the narration the quality of an intrigue,
constitute a practical difference between the experimental sciences and the field sciences. The latter, as I have characterized
them in The Invention of Modern Science, construct stories in the
sense that the causes they present can no longer claim to have
the power to determine how they cause. The question "What can
the cause cause?" here assumes an importance that is foreign
not only to the cause associated with the Galilean object, which
provides the = sign with its power, but also to the causes associated with all the practices of staging and delegation common
to experimentation. Operations such as staging or delegating
assume stable relationships and roles, which is precisely what
the "field" challenges and for which it substitutes the interest
of intrigue.

THE

EMERGENCE

PRACTICES

OF

EMERGENCE

231

The appearance of scientists endowed with the appetite I


have just described is an important ecological fact within the
population of contemporaiy scientific skills, but the meaning
this fact may harbor depends on that ecology. For a long time,
"Darwinian" science has been presented in a form that enabled
it to claim the same power to judge as the laboratoiy sciences.
Natural selection had to be all-powerful so that its representative could claim the power to judge, to explain, even rhetorically,
the histoiy of living creatures. That we would arrive at the "just
so stories" of sociobiology when speaking of primate or human
behavior was, in this sense, entirely predictable. What is much
more interesting is that some Darwinian biologists today seem
capable of presenting themselves differently. I am referring to
Wonderful Life: The Burgess Shale and the Nature ofHistojj, in which
Stephen J. Gould states that the field sciences are now capable of
claiming the uniqueness of their practice, of inventing themselves as different without fearing the judgment that would call
them inferior.' 4.
Moreover, although the new EcoDevo (ecological developmental) biology explores the embiyo's development with the full
array of sophisticated tools provided by experimental science,
it is nonetheless something like a "field science." The field in
this case is the amazing "causal choreography" associated with
processes of development that had been characterized by both
finalists and neo-Darwinists as directed by a cause (the final
cause or the program). The characterization of the continuously self-redefining developmental entanglement mobilizes
all the words we have to describe encounters that affect the veiy
fate of the encountering terms. From infection or mobilization
to hijacking, seduction or reciprocal induction, the common
feature of these narratives is that any simple relation between
"cause" and effect" is lost without regret.
When the interest I associate with the field sciences is
addressed to living organisms customarily judged in terms

222

T H E P R A C T I C E S OF

THE

EMERGENCE

P R A C T I C E S OF E M E R G E N C E

233

of purpose, as creatures of natural selection, directed toward

determine or master, has given rise to new values, new inter-

an end, what is learned instead is the risk of such judgment, a

ests, and, of course, new claims.

risk that cannot be overcome, that will recur at each successive

Thus, new, practical faces of emergence, resulting from the

step. Such sciences not only speak about the hazard of circum-

experimental sciences as well as from the sciences of artifice,

stance, they create interest in the intrigue that binds heteroge-

arise, which will allow us to explore possibilities of articulation

neous elements whose meaning is produced in the encounter

between laboratory creations and field creations. Some may

itself. In doing so, they serve as a decisive ingredient in the

criticize these new faces as masks that conceal a new strategy for

problem of emergence. For, the two confrontational positions

conquering the terrain. Indeed, self-organization can be seen as

that destroyed this problem are similarly challenged. Neither

a new "all-terrain" response. In fact, it was the clearly differen-

the finalist biologist, for whom the ends of organization define

tiatedenthusiastic or disparagingbut all too often caricatured

the irreducibility of emergence, nor the "reductionist," who

responses engendered by nonequilibrium physics that forced

accepts his adversaries' purposes as such only to relate them to

me to take the first steps toward what I have here referred to as

the power of selection, have any desire to conceptualize the dual

the "ecology of practices."

indeterminacy of "ends" and "means." The question that now

These reactions also indicate the limits of the "interdisci-

arises concerns practices that would eagerly welcome this dual

plinary project" of which Order out of Chaos: Man's New Dialogue

indeterminacy, practices that would require an alliance with the

with Nature, which I coauthored with Ilya Prigogine, was a part.

field sciences in order to construct the problem of articulation

Whenever the question of scientific practices is involved, inter-

between the requisites of emergence and the purpose that will

disciplinarity, whether it finds the source of its references in

be associated with what emerges.

physics or cybernetics, information theory or some "theory

To approach this question, I want to examine the answers

of complexity," suffers from the same weakness as the con-

supplied by the experimental sciences and, more specifically,

cept of an "idea" (or an ecology of ideas). The idea seeks to "be

those sciences that, during the past years, have claimed to

applied" and is eager to exaggerate any resemblances. It entails

"renew" the question of emergence: the physical chemistry of

no requirements or obligations, and therefore travels freely as

nonequilibrium and the study of neoconnectionist networks.

some kind of shared currency that would permit an "exchange"

I will try to show that, in both cases, a mutation is produced

or "dialogue" among the sciences but that dissimulates the glar-

with respect to the domain of origin. The physical-chemical

ing difference among the use-values it is able to claim in various

being "far from equilibrium" may cause a divergence between

scientific domains. So, it is not in terms of "interdisciplinary

"condition" and "determination" whose coincidence was for-

promise" that I conceive of the possible faces of self-organiza-

merly ensured by the state of thermodynamic equilibrium. The

tion, but in terms of the test I associated with emergence as a

artificial "neoconnectionist" being brings about a divergence

problem: a practice of articulation that brings about and stabi-

between "fabrication" and "mastery," which the watchmaker's

lizes abandonment of the position of a judge who has no need of

artifice celebrated. Such divergences are what the term "self-

a terrain because he knows ahead of time what that terrain has to

organization," shared by both domains, reflects. For the sci-

say. Whenever there is a question of emergence, indeterminacy

entists who suggested it, the loss of power, that of the ability to

must become a part of the meaning of what is constructed in the


laboratory.

DISSIPATIVE

COHERENCE

235

heat" that expresses the fact that any deviation from the ideal
results in "dissipation." And when the problem is not displace-

15

ment, ideal or not, from one equilibrium state to another, but


an evolution toward equilibrium, only the maximum value of

Dissipative Coherence

the entropy is defined, corresponding to the equilibrium state.


The evolution of an isolated system toward equilibrium "causes"
undefined entropy to increase until it reaches its well-defined
maximum.
The definition of entropy, and other thermodynamic potentials, thus gives a central role to the concept of an equilibrium
state. More specifically, the two concepts define each other: the
equilibrium state is defined by the maximum or minimum value
of the potential (according to the definition of this potential)

In the next few pages, I want to return to physics, but not the

and the potential guarantees the stability of this state. Once at

physics of laws. The physical chemistry of nonequilibrium

equilibrium, the system remains there and any evolution that

refers, through "equilibrium," to thermodynamics, a "phenom-

would spontaneously move the system away would contravene

enological" physics that was said to have been reduced to the

the second law of thermodynamics. For example, in an isolated

terms of the probabilistic interpretation that led to the triumph

system, it would correspond to a decrease in entropy. The exis-

of the laws of the Queen of Heaven (see Cosmopolitics, Book III).

tence of a thermodynamic potential function thus characterizes

I want first to briefly recall the rather curious structure of

a dissipative evolution by its final state, when all dissipation will

so-called equilibrium thermodynamics characterized in Book

have vanished. In short, the thermodynamics of equilibrium is

III. This science stands out in that its object is not energy, or

by and large characterized by the opposite of the obligations of a

thermodynamic, processes as such but their "rational mim-

field-based approach: its questions revolve around a state that

icry": the displacement of equilibrium, where process time is

is unique precisely because it has the power to silence all ques-

replaced by the progressive manipulation that forces the transi-

tions, that is, to provide a final reckoning for a process for which

tion from one equilibrium state to another infmitesimally close.

it has neither the means nor the need to give an account.

In this way, we arrive at the three-part definition of entropy as a

In Cosmopolitics, Book V, I introduced Ilya Prigogine as the

state function. In the ideal case, when the change of state it mea-

successor, then as the heir, to Boltzmann. However, when the

sures is a reversible displacement between equilibrium states,

work that resulted in his 1977 Nobel Prize is being discussed,

entropy is defined in terms of the system variables. When this

he should be referred to as a student of Theophile de Donder.

displacement does not fully satisfy the ideal of a transforma-

Successor and heir are a matter of choice, being a student is pri-

tion that never brings the system at a finite distance from equi-

marily a "fact," even if this fact also implies a choice (not every

librium, entropy remains a state function, but its definition

teacher becomes a "master" for her students). De Donder was

becomes indeterminate: it includes some "uncompensated

a mathematical physicist and a correspondent of Einstein. For

234

DISSIPATIVE

23 6

COHERENCE

him, science was something that "embodies the purest image


that the sight of Nature can bring to life in the human mind." And

DISSIPATIVE

COHERENCE

337

the dissipative activity of matter.


With de Donder, thermodynamics, a science that is delib-

when he was required, out of professional duty, to teach thermo-

erately blind to what it cannot subject to a rational equivalence

dynamics, he did not find that purity. So he decided to create it.

through which it can articulate its variables, would reorient

For that to happen, Clausius's mute, uncompensated heat would

itself around a new physical-mathematical being that, in itself,

have to learn to speak, would have to participate in the harmony

said no more than other thermodynamic properties but raised

of functions and reveal the musical truth of the indistinct noise

a question where those other beings gave only answers: it is the

known as dissipative evolution. And de Donder turned to that

production of entropy that describes the growth of "uncompen-

field of thermodynamics where dissipation is entirely intrinsic,

sated heat" over time. Concerning the production of entropy,

where the ideal of a reversible transition from state to state is

thermodynamics as such doesn't say much, except that it is the

the most obviously artificialchemistry. For, measurement by

most general of thermodynamic potentials. Regardless of the

means of reversible displacement was able to normalize chemi-

conditions that define a system (isolated, constant temperature

cal reactions only by stripping them of their most important

and pressure, etc.), the production of entropy at equilibrium

characteristics: the spontaneous heat given off or absorbed by

is, by definition, identically zero. Correlatively, all irreversible

eveiy reaction and the reaction rates that qualify them and that

evolutions toward equilibrium are, by definition, "entropy pro-

kinetics studies.

ducing." The enigma has become a problem: with what kind of

Dissipation and chemistry. It required the freedom of a


mathematical physicist inhabited by the beauty of his science to

variable can this production of entropy be associated?


Chemistry is privileged in that, aside from thermodynamic

challenge the hierarchical structure that sanctified the power of

variables, chemical transformations are characterized by other

that science. De Donder did this in two ways: by asking about

variables that immediately introduce time: the kinetic vari-

irreversibility, which had been deprived of any meaning on

ables that refer to equilibrium as the state in which processes

the fundamental level, and by attributing to it, as its "topos,"

continue to occur but with velocities such that their effects are

as the site where the corresponding problem could be con-

canceled. De Donder would make the production of entropy the

structed, a chemical activity that had been reduced to the inter-

setting in which an explicit link could be forged between irre-

action between the atoms of physics. In discussing the growth of

versibility, process, and time. Forging refers to the art of making

entropy, I spoke of an "enigmatic factish" that raises questions

alloys, which force disparate materials to become one. In this

it is, as such, incapable of answering (see Cosmopolitics, Book

case, kinetics, which evaluates chemical velocities, and thermo-

III). But the enigma in question cannot be separated from the

dynamics, which qualifies chemical reactions on the basis of

final decades of the nineteenth-centuiy crisis concerning the

the equilibrium state in which the production of entropy is can-

values and obligations of physics. For de Donder, who was in the

celed, although previously rivals, will now be forced to partici-

service of harmonious beauty rather than the power to impose

pate in the definition of the production of entropy.

requirements, the crisis never existed. The enigma was free to

The production of entropy is the crucible where the link

become a question, and that question created an interest in what

is forged, in the sense that it defines the question asked by de

it designated as the terrain on which it could become a problem:

Donder of eveiy chemical reaction: What do you contribute to

D I S SI P A T I V E

COHERENCE

the production of entropy? Here, the term "contribution" is


critical. It seems to refer to the possibility of "judging" a chemical reaction on the basis of a "value," the value of its contribution. Earlier, Clausius had judged the respective values of the
conversion of heat into work and the passage of heat from one
temperature to another using as a fixed point their equivalence
defined by the ideal Garnot cycle. When the system has returned
to its initial state, conversion and passage are exactly balanced.
But the "value" defined by de Donder (in this case, the contribution of each reaction to the production of entropy is defined by
the product,

of its thermodynamic potential, the affinity A,

and its velocity v) is part of the question, not part of the solution. The overall production of entropy must be positive, yes,
but this thermodynamic definition leaves the specific contribution of each chemical reaction to this production indeterminate. This allows de Donder's thermodynamic^, which defines
the overall production of entropy for the ensemble of reactions
that participate in a chemical transformation as being positive
everywhere, except at equilibrium, where it is zero, to take into
account kinetic description, which describes this ensemble as
a "system" of reactions that are coupled to one another. From
the point of view of kinetics, each reaction has a velocity that
depends on the concentration of its reagents, that is, on the
other reactions that contribute to the production or destruction
of the reagents in question. And this coupling of reactions can
result in the fact that some of them provide a negative contribution to the production of entropy.
The production of entropy, therefore, can be used to present a problemthe difference between each separate reaction and the ensemble of coupled reactions. And this problem
is open-ended. The thermodynamic condition expressing the
second law, the positive production of entropy, is inadequate
for determining its solution. Except in one case. At equilibrium, it is thermodynamics's glorious simplicity that triumphs:

DISSIPATIVE

COHERENCE

239

the rate of each reaction is separately compensated by the rate


of the inverse reaction. However, the production of entropy
as de Donder redefined it only makes the significance of this
simplicity explicit. At thermodynamic equilibrium, the coupling of dissipative processeskinetic descriptionis without
consequence. But the clarification, as is the case whenever it
is creative, transforms the meaning of what it makes explicit.
Equilibrium is no longer the thermodynamic "state," but a particular situation within a landscape that asks to be explored. And
if the system is maintained out of equilibrium? If, instead of letting it evolve toward the state that has the power to make coupling insignificant, the experimenter forces exchanges between
the system and the exteriorfor example, through the permanent flow of chemical reagents that prevents it from reaching equilibrium? What might then be the effect of the negative
contributions to the production of entropy that makes coupling
between processes possible?
This was the problem Prigogine set for himself, and he provided it with a clear purpose from the outset. For Prigogine, a
student of de Donder, was preoccupied with the question of
emergence as exhibited by biology. And the uniqueness of his
position among physicists arises from the fact that he wanted
physics to have the ability to address this question. Living
organisms do not have to be answerable to physics, it is physics that has to be answerable to the fact that the living organism is actually possible. In other words, Prigogine required that
physical-chemical processes become a relevant "terrain" for
the question of life. He insisted that "irreversibility"the production of entropybe able to tie its fate not with the evolution
toward equilibrium but with the processes that, in one way or
another, constitute a living organism.
In his doctoral dissertation, Prigogine generalized to all
physical-chemical processes the relationship that de Donder
had forged between chemical kinetics and the production of

D I S SI P A T I V E

COHERENCE

DISSIPATIVE

COHERENCE

241

entropy. In 1945, he showed that if exchanges with the envi-

disorder. And Schrodinger, celebrated as a precursor by molec-

ronment constrain the system to remain outside equilibrium

ular biologists, assumed that it was the chromosome (DNA had

but maintain it close to equilibrium, the evolution will reach a

not yet been discovered) that contained and transmitted the

stationary state (non-zero velocities) determined by the mini-

secret of this order, an order that was defined, given that it was a

mum (non-zero) value of the production of entropy compatible

question of struggling against physical irreversibility, by the lan-

with those exchanges. In this way, the power of thermodynamics can be extended to the neighborhood of equilibrium. And in
collaboration with the biologist jean Wiame, Prigogine immediately published an article in which he examined the possible
relevant relationships between his theorem and the question
1

of the living organism. In it, he showed how the stability of the


stationary state when entropy production is at a minimum may
be associated with various properties of the organism.
The limited scope of his theorem was not an obstacle to
Prigogine, for he had established a critical finding: the equilibrium state has become a special case (where the production of
entropy is zero because the constraints are zero) and stationary states close to equilibrium can be characterized by a certain
"order." For instance, spatial differentiation of the concentra-

guage of artifice: chromosomes contain both the law and the key
to the means for implementing that law, they must explain both
the "program" the living organism obeys and the mechanisms
that give that program the power to direct the development and
operation of the organism. But for Prigogine, the living organism was dissipative. It did not have to maintain itself against
entropic disorder, it challenged the simple identification of
the increase of entropy with disorder. For, quite obviously, the
living organism causes entropy to increase. In order to live, it
must feed itself. To extend Schrodinger's concept of negentropy,
we could say that for Prigogine it is the processes that produce
entropy that we must turn to in seeking the key to those "negative contributions" required by the order characterizing the living organism.

tion of chemicals can appear in a system subject to a continuous

Yet, it was only in 1969 that the now famous term "dissipa-

temperature differential: thermodiffusion "couples" the ther-

tive structure," expressing the association of order and dissi-

mal diffusion that produces entropy to the chemical "antidiffu-

pation, came into use. It celebrates Prigogine's new assurance

sion" resulting in negative entropy production, which would be

that he had resolved the contradiction between entropic dis-

impossible in isolation. Isn't the order associated with the living


organism also impossible when that organism is cut off from its
exchanges with the environment?
During this time, Erwin Schrodinger published his highly
celebrated What Is Life?. The contrast between the two "physicalist" approaches to the order that characterizes the living
organism is revealing. For Schrodinger, the order of the living

sipation and the emergence of order required by living organisms. A crucial element of his work, which extended over more
than twenty years and for which he was awarded the Nobel Prize,
is the divergence between "condition" and "determination"
required by certain couplings of dissipative processes far from
equilibrium.
The production of entropy, which is minimal near equilib-

organism is "negentropic," characterized by "negative entropy,"

rium, has the power of thermodynamic potentials. Based on the

which implies that this order questions, in one way or another,

conditions it determinesthe intensity of the flux that imposes

the principle of entropy increase. The living organism imposes

a fixed distance from equilibriumits minimum value can be

the concept of an order capable of resisting dissipation and

used to determine the stationary state and ensure its stability.

24,125 D I S S I P A T I V E

COHERENCE

DISSIPATIVE

C O H E R E N C E 247

And it is this thermodynamic power that Prigogine initially

fluctuation" indicates that the practice of physicists has diaiifii .1

attempted to extend far from equilibrium, until his work, con-

If they are no longer able to require that a potential limelinn

ducted in collaboration with Paul Glansdorff, underwent a prac-

be defined, they are obligated to address the problem of

tical mutation. For the interest in far-from-equilibrium states

sible instability. They are obligated to "test" an activity regime

came to be associated with the fact that they escape the power

deduced from its equations in order to determine if that regime

ptiH

expressed by the possibility of defining a potential, or a state

will be restored when subjected to a perturbation or, on the con

function: identifying the end point of dissipative evolution and

traiy, if the perturbation will increase. The perturbation intro-

ensuring the stability of that final state. The central question

duced by the test expresses the question imposed by an activity

instead became that of stability or instability, and it is the cou-

regime that must be conceived as intrinsically fluctuating, once

pling of processes that the answer to this question depends on.

the insignificance of those fluctuations can no longer be guar-

In other words, process coupling no longer serves only to define

anteed. That is why, in the case of instability, the physicist can

the production of entropy but replaces the production of entropy

describe a fluctuation as what will be amplified and the new, sta-

as the focus of the definition of the activity regime toward which

ble activity regime as a "giant fluctuation" stabilized by irrevers-

the system will evolve. In short, far from equilibrium, the states

ible processes. But "chance," here, has no value independent of

that prolong the stationaiy state near equilibrium ("thermo-

the (nonlinear) coupling that creates the landscape of possibles

dynamic branch") may become unstable. "Dissipative structure" is the name given by Prigogine to the new activity regimes,
which owe their stability to interprocess coupling.

and the question: which will be realized? 3


To speak of an "activity regime" as I have done expresses the
fact that it is no longer possible to speak of a "state," for the defi-

The remainder of this history is outside the scope of this

nition of a state always follows from a power relationship, the

book. But I want to discuss two aspects that help clarify the prac-

satisfaction of the requirement that the definition of a system's

tical novelty of Prigogine's approach. One is the associationthe

identity means the ability to define its state(s). But this power

source of several fads and misunderstandingsbetween dissi-

relationship also disappeared along with the disappearance of

pative structures and "order through fluctuation."'-' The other is

thermodynamic potential. Whatever controls the external vari-

the term I introduced above, "activity regime."


Order through fluctuation expresses, in an immediate way,
the emotive charge of the event for a specialist in thermodynamics: the loss of power for the second law, which, through
a potential function, ensured the stability of the state, that is,
the regression of the inevitable and unceasing "fluctuations"
(the state being described in terms of macroscopic values, or
means). The fads and misunderstandings arose because the
concept of fluctuation was associated with a "cause" or "responsibility": chance fluctuations would be responsible for a "choice"
or else would be creative. Rather, the expression "order through

ables (pressure, temperature, reagent flow) no longer controls


the system. Not only does the second law of thermodynamics
no longer guarantee that uncontrolled local fluctuations will
regress without consequence, the very identity of the system can
be transformed. A factor that is insignificant at equilibrium,
such as the existence of the gravitational field, can play a crucial
role, that is, it can make distinct activity regimes possible. This
"sensitivity" of the activity regime when far from equilibrium
to factors that are insignificant at equilibrium transforms the
nature of the questions asked. For now it is the activity regime
of the system that determines its relevant definition, what this

244

DI S SI P A T I V E

D I S S I PAT I V E

COHERENCE

definition must take into account, what the system can become
"sensitive" to. Therefore, to study an activity regime is also to
study the stability or instability of a definition this regime might
justify under certain circumstances but might cause to be modified in others. Correlatively, the notion of "constraint" assumes
a meaning very different from "limit." Relationships with the
"exterior" constrain the system to remain far from equilibrium,
but the "limit conditions" do not provide the ability to determine
(as was the case near equilibrium) what, from among the various
possibles, will eventually be realized. Although limits are usually
associated with meaningful imposition, far-from-equilibrium
constraints are given their meaning by the activity regime they
make possible. And this meaning will be determined by the production of the solution to the problem posed by the constraint.
In any event, the constraint will be a condition, but it will lose
the ability to determine what it might be a condition for.
Although physical chemists have not abandoned the concept
of a system as such, that concept no longer corresponds to the
power relationship resulting from the ability to deduce possible behaviors from the definition of the system. They preserve
the concept because they can. What they address has been prepared in the laboratoiy, and they know what the definition of a
system at equilibrium allows them to treat as negligible. That is
why they can trace the landscape in which stable and unstable
activity regimes can be distinguished, and the bifurcations that
indicate a transformation of distribution between stable and
unstable. But this ability to define the landscape of possibles
as preexisting the realization of a given possible is now strictly
correlated to the power given to the physicist by preparation in
the lab. The concept of an activity regime as such entails a distinction between the "abstract" problem, expressed in terms of
constraints, and the concrete solution produced by the effective coupling of processes in space and time. It is this distinction that has been, biographically, my pathway to the Deleuzian

COHERENCE

245

distinction between actualization and realization, or the virtual


and the possible.
Under far-from-equilibrium conditions, the scientist can
no longer "require" but the activity regime can "obligate." It is
this new configuration of requirements and obligations that is
referred to by the term "self-organization." The term was, significantly, borrowed by Prigogine from the tradition of prewar
"antireductionist" embryologists (notably Paul Weiss), who
noted the ability of the embryo to determine for itself what would
be a ' cause" and what would be insignificant. It tells us that
physicists who adopt it now consider themselves to be directly
confronting the problem of emergence: far-from-equilibrium
physics is able to "comprehend" the arguments of embryologists against the reductions that had, in one way or another,
joined forces with the defining power of a state. However, the
term "comprehend" has two distinct meaningsto include and
to understandand only the distinction between the two can
create the difference between a physics that claims to "explain"
emergence and a physics that becomes a partner in a practice of
negotiation through which emergence can be constructed as a
problem.
Chemical clocks that exhibit periodic behavior, Benard cells
that imply coherent collective movement at the macroscopic
level in a crowd of innumerable moleculesfar-from-equilibrium physics results in the creation of new experimental factishes that signal the emergence of activity regimes that break
with the general ideas associated with microscopic disorder.
Here, laboratory and theory work in tandem, constructing new
descriptions about emerging "order" and natural processes
that can be included within the new framework. However, this
inclusion concerns beings who "depend" on a "constraining"
environment, but, unlike living organisms, have not created a
milieu for themselves. It is true that far from equilibrium, coupling, or the interrelation of processes that produce it "create"

24,6

DISSIPATIVE

COHERENCE

a being whose behavior cannot be identified as the effect of


the constraints imposed by its environment. But the being in

DISSIPATIVE

COHERENCE

247

which they can encounter as such only if they first recognize it as


material for obligation.

question appears or disappears, depending on circumstances,

The promising factish, therefore, creates interest in the

without a fuss. Additionally, the coupling between processes is

"terrain" where a promise and its recipient will be actualized.

silent about differences that may matter if a "purpose" comes to

But should we take seriously the intense association between

be involved; for instance, the difference between two possible

the promise and a face-to-face situation, when looking into

activity regimes, one chaotic and the other periodic. Physical-

the face of the one who promises? Yes, possibly, providing that

chemical self-organization is "factual" in nature, and the term

"faciality" is understood in the asignifying, asubjective sense

"organization" here does not correspond to any kind of "pur-

given by Deleuze and Guattari. The face is not that of the "other

pose" that might articulate, for anyone but the scientist, risks,

who promises," the foundation of intersubjective relationships,

values, and challenges.


Yet, those same beings are also factishes of a veiy differ-

it is an "abstract" machine, inductive of deterritorialization, and


it is as such that it can, in certain circumstances, become a con-

ent kind, one that does not bear witness. For a long time I have

dition of the signifier and the subject. Face as "voice carriers"

searched for an adjective that would reflect this difference and,

are not, for Deleuze and Guattari, an anthropological universal:

finally, it is the word promising that seems most appropriate,

"These are very specific assemblages of power that impose sig-

because it binds ordinarily disparate semantic uses. What is

nifiance and subjectification." 4 Similarly, here, there is a veiy

"promising" often refers to a self-interested approach, whereas

specific appetite that the "promising factish" stimulates when it

a promise is often associated with a commitment. The "techni-

induces a new kind of relationship between the laboratory and

cal" innovator I have presented as a practitioner of invention

the world. The question of what a being "far-from-equilibrium"

most often assumes, as her point of departure, a "promising"

is capable of suddenly refers to the price paid for the power of

possible, only to discover that there is a great distance between

the laboratoiy. For, it is "outside" where the answers become .

what she thought was promised and the actualization of that

more interesting, where the power of the laboratoiy is replaced

promise. In fact, what is "promising" promises nothing in par-

by the possibility of "reading" histories, of "following" the part

ticular to anyone in particular. Unlike the promise, it has no

played by an activity regime whose intelligibility has been pro-

recipient but stimulates, on the part of whoever allows himself

duced by laboratoiy practices within those histories.

or herself to be captured, the appetite (quite unlike that of the

If physical-chemical self-organization is a fact, it also raises

experimenter) I have associated with the verb envisage. Those

questions that the "fact" is unable to answer but for which it is

who envisage on the basis of a "promising" problem or possibil-

available as something "promising." A self-organized activity

ity know they are obligated by a world, even if they don't know

regime could indeed be what the emergence of living organisms

how that world obligates them. They know that actualization of

requires. What is requisite and not what explainsthis distinc-

what they envisage as promising implies a creation of meaning

tion reflects the difference between understand and include,

about which they cannot freely decide. Unlike the experimenter,

both of which the verb comprehend refers to. The requisite refers

they negotiate with a world they must encounter as partly inde-

to what a problem needs, without which it could not be pre-

terminate, susceptible to new relationships of meaning, but

sented. It implies that the scientist situates herself with respect

24,128

DISSIPATIVE

COHERENCE

DISSIPATIVE

COHERENCE

247

to the living organism as she would when confronting a prob-

is relative to the pertinence of the knowledge they will capture

lem, rather than a "fact," for example, the fact constituted by the

and rearrange. But with them the authority of physics moves

ensemble of far-from-equilibrium activity regimes in a phys-

forward, for it alone, situated at the summit of the hierarchy of

ical-chemical system. If the possibility of such regimes desig-

the sciences, is authorized to determine what has a right to truly

nates what is requisite, this means that they invite us to wonder

exist, and what must be categorized as an illusion.

about the meaning they might have assumed within a stoiy they

Order out of Chaos was inhabited by the hope of a new coher-

do not explain. They "promise" in the sense that they can be

ence in our knowledge, one that would heal the deep rifts created

used to identify an issue: what must be "narrated" is the way an

by physics's denial of becoming. It called for a dialogue among

activity regime far-from-equilibrium eventually came to play a

the sciences, united by the open question of becoming, a ques-

role, that is to say, assume a meaning for something other than

tion none of them could appropriate. 6 What this well-inten-

the scientist. In this case, and from the point of view of this role,

tioned offer neglected is that those sciences were also modern,

not all far-from-equilibrium regimes are equivalent. A stoiy

and haunted by the power of disqualification and conquest,

could be told in which the stabilization of one regime rather

more eager for an alliance with the generalizing power of the

than anotherbut also, possibly, mutated assemblages, trans-

"new model" revealed by physics than for the risks, questions,

formations of couplingcould assume a meaning the laboratory

and challenges the physicist's "desire" might arouse. At best, in

cannot provide, the meaning of events affecting a "body," that is,

practices such as biochemistry or the study of "social insects,"

events capable of an evaluation that implies and initiates a dis-

which already dealt with the contrast between multiply coupled

tribution between "interior" and "exterior," between "function"

activities and coherent overall behavior, connections were made

and "milieu." The requisite creates an appetite for situations in

without too much fanfare. I'll return to this later. But at worst,

which the "fact" would become an issue for something other than

and with considerable commotion, new paradigms (with short

the scientist. It creates an appetite for the problem of emergence,

life spans) were announced, turning dissipative structures or

for stories in which the question of the "value" of the possibles it

order through fluctuation into one of those all-purpose concepts

allows would be invented. 5

that seems to proliferate wherever the will to science takes the

It is this new appetite, drawing the physicist "outside the


laboratory," that the original title of Order out of Chaos, The New
Alliance, expressed. The physicist evoked by the "new alliance"
would no longer be interested solely in the "world" she has
learned to judge in the laboratory, it is the diversity of "cases"
she desires. She must then form working alliances with the
diversity of knowledge practices liable to identify couplings,
arrangements, and coherent collective behaviors whose meaning would "emerge" from asignifying local activities. But here
too the question of an ecology of practices is relevant. Physicists
may indeed limit themselves to proposing models, whose value

place of practice. The "promising factishes" invented by physics


can become ingredients of a practice of emergence only if the
world they encourage us to investigate is populated with knowledge that is not awaiting the surplus legitimacy and power that
physics-inspired "model making" would confer upon it. Such
knowledge would need to be capable of obligating the physicist
to take an interest in what the model must capture in order to
actualize the promise and its recipient.

ARTIFICE

AND

LIFE

I want here to turn my attention to the field of "artificial life."


The promoters of this field associate its ambitions with what

16

Artifi.ce and Life

they define as the failure of "artificial intelligence." It is not possible to construct a "brain" capable not only of reasoning but,
especially, of learning to explore a milieu and extract from it the
ingredients of "adapted" behavior, unless we have first endowed
that brain with a computing "body" capable of encountering that
milieu, of moving, falling, touching, and taking into account
the consequences of its own actions. Which is to say, unless we
have raised the question of the "evolution" either of a population of such bodies or the behaviors characterizing that body.
We must first construct a body one can call living, one that is
capable of learning, before we can construct a being we could

The problem of emergence in the context of the sciences of


the artificial presents itself quite differently. Here, the question is not at all that of the possible "emergence" of meaning
"for" a world. The artifact always has meaning. It can always
be understood in terms of a logic that relates means to ends.
If we consider the two distinct fields, "artificial intelligence"
and "artificial life," that, in the past thirty years, have claimed
to "explain" intelligence or life, the common trait that characterizes them (and expresses their shared connection to John
von Neumann's work on computers) is the radical distinction
between "information processing," which must be understood
in logical terms (computation), and the material "implementation" of such processing. Here, the concept of an artifact should
no longer evoke the image of a clock, nor that of a robot laboriously assembled. What human art intends to reproduce is the
"form" that controls matter, that is, which can be conceived
independently of the matter it will control. Computer beings are
not actualized; they are indifferently realized for a given physical
medium. This physical medium can be a source of breakdowns
or crashes, but not differentiation. Here, the artifact is staged
in a way that is foreign to the natural sciences: "mind" controls
matter. The ideal is submission not to laws but to a project.
25

call thinking.
It is quite remarkable that the physical-chemical self-organization I have presented and the "artificial" self-organization
I am about to present converge from opposite horizons toward
the question of the "body," a being endowed with a topology that
creates a substantive difference between "interior" and "exterior," to which corresponds a differentiation between two types
of "variables."' The body forces a distinction among variables
that refer, to return to Feibleman, to the "level below"variables that, if they belong to a body, no longer characterize physical-chemical interactions but relationships that have a meaning
"for" the bodyand those that refer to the "level above," which
correspond to the milieu that exists for the body and for which it
exists, a milieu of welcome or catastrophic encounters, a milieu
in which not eveiything has the same value from the point of
view of the risky wager that has produced a given body.
The two horizons are indeed opposed. The "promising factishes" of physical chemistiy pose the problem of the emergence
of an "exterior" that serves as a milieu. In contrast, if there are to
be "promising factishes" created by artificial life, the definition
of a "milieu" would not pose a problem. On the contrary, it is the
value "for the exterior" that generally fully defines the artifact.

252

ARTIFICE

AND

LIFE

ARTIFICE

AND

LIFE

53

Here, the problem will be one of the emergence of variables that

In order for the bet to pay off, computer scientists or robot

can be called "internal," and which must not be defined from

manufacturers, for example, would have to agree to identify

the viewpoint of an "external" finality. In other words, the term

their products as part of this field, to refer to it, to situate them-

"self-organization," shared by both fields, does not have the

selves within the perspective it promotes. And to do that, they

same meaning. In physical chemistry, the "autonomous" charac-

would have to see some benefit in it. Their products, when situ-

ter alluded to by the prefix "self - " is something acquired, but the

ated within the framework of emergence, of the manufacture of

possibility of speaking of "organization" has yet to occur. In the

life, of a contribution to its logical identification, would have to

case of "artificial life," organization is something acquired, but

become more interesting than if they were situated within the

the possibility of characterizing it as autonomous is in question.

more traditional framework of technological innovation. How-

The field currently known as "artificial life" may have


claimed that the ambitions of its predecessorthe emergence
of intelligencewere very premature, but its own ambitions
are not much more modest. Chris Langton, a leader in the field,
wrote the first manifesto for the inaugural conference in Los
Alamos in September 1987. Eveiy word was carefully weighed
and eveiy visionary accent carefully deliberated: "Artificial life
is the study of artificial systems that exhibit behavior characteristic of natural living systems. It is the quest to explain life in any
of its possible manifestations, without restriction to the particular examples that have evolved on earth. This includes biological
and chemical experiments, computer simulations, and purely
theoretical endeavors. Processes occurring on molecular, social

ever, this was not quite the case. It is quite probable that these
products will confront "humanity," as Langton writes, with
a number of far-reaching challenges. Yet, how can we fail to
recognize, in the way in which he presents those challenges, a
mobilization, in the furtherance of scientific ambition, of the
ancient figure of "man" defying the order of creationman's
confrontation with the product of a knowledge that would finally
fulfill his ultimate goal, the definition and reproduction of "life"
as such, independent of the contingency of his earthly origins?
Knowledge products do create, and will, of course, continue to
create, challenges, although more dispersed, arising from the
labyrinth of technological innovations that capture and reinvent
for their own use what Langton wishes to mobilize.

and evolutionary scales are subject to investigation. The ultimate

However, "artificial life" does not simply satisfy a mobilizing

goal is to extract the logical form of living systems. Microelec-

rhetoric. Something has happened, a "factish" has been invented

tronic technology and genetic engineering will soon give us the

and recognized that has created the possibility for a relative

capability to create new life forms in silico as well as in vitro. This

mutation of what we understand by an artifact. While the scope

capacity will present humanity with the mostfar-reachingtech-

of this event will not be what Langton hoped for, nonetheless,

nical, theoretical, and ethical challenges it has ever confronted.

it may enable emergence to partly escape the traditional frame

The time seems appropriate for a gathering of those involved in

of the "human artifact/living organism" analogy. But to address

attempts to simulate or synthesize aspects of living systems." 3

this issue, it is best to abandon Langton, who, from his comput-

Some twenty years later, it cannot be said that "artificial life"

er's keyboard, wished to be the creator of "worlds" populated by

has kept the prophetic promises of its promoter. Artificial life

quasi-livingcreatures. 3 It is not a matter of judging the scientific

was a gamble because its federative ambition depended on its

value of a work but of turning away from the slightly premature

ability to mobilize the various fields enumerated by Langton.

questions it has inspired: "What of man's view of himself? He

252

ARTIFICE

AND

ARTIFICE

LIFE

AND

LIFE

53

now takes pride in his uniqueness. How will he adjust to being

associated with the order of living organisms must reject the

just an example of the generic class 'intelligent creature'? On

triumph of someone who succeeds in getting his artifact to do

the other hand, the concept of 'God' may take as much a beating

what he wanted it to do. By way of affirming a value that refers

as the notion of 'man.' After all, He is special now because He

to the type of order that would be needed to characterize "arti-

created us. If we create another race of beings, then are we not

ficial life," Kauffman relates a requirement and an obligation

ourselves, in some similar sense, gods?" 4 We need to abandon

that question the possibility of referring to the creator's project

the apparent grandiosity of such claims for they conform to a

as bearing exclusive responsibility for the artifact's creation. As

mythic mold that has been reduced to cliche. I now want to turn

for the value he affirms, we still need to examine the two terms

to the man without whom I would never have investigated the

used enigmatically by Kauffman to characterize it: "generic" and

question of artificial life, Stuart Kauffman.

"heart."

For Kauffman as well, God is not far away. This is how he

In mathematics, the term "generic" designates a behavior

recounts the passion that has driven him ever since he began

that is not only "robust" in the sense of being relatively stable

tiyingto understand life: "I've always wanted the order one finds

compared to perturbations or the imprecision of initial condi-

in the world not to be particular, peculiar, odd or contrivedI

tions.'" The property of genericness implies that behavior is also

want it to be, in the mathematician's sense, generic. Typical.

qualitatively stable, in terms of the details of the relations, con-

Natural. Fundamental. Inevitable. Godlike. That's it. It's God's

nections, and interactions that bring it into existence. We can say,

heart, not his twiddling fingers, that I've always in some sense

trivially, that evolution toward equilibrium is a form of generic

wanted to see." 5

behavior for physical-chemical systems because it may be

So, Kauffman wants to "see" order as godlike and not


"become god" as Langton did. Also, he doesn't want to see
"God's twiddling fingers," which are the required intermediaiy in the clock metaphor of creation. According to the meta-

characterized by the diminishing significance of interprocess


coupling. But the term can only be used for equilibrium retroactively, following the discoveiy of much more unexpected kinds
of generic behavior.

phor, we cannot identify with God in terms of his ends, which

Kauffman himself participated in the early history of the

for a believer are impenetrable; but we can recognize the work

field. In 1965, as a young student already excited by the themes

of his "fingers," the arrangement he has imposed upon mat-

of complexity and self-organization (in the tradition of "sec-

ter. Particular, odd, contrivedthese are the adjectives that

ond-order cybernetics" associated with the names of W. Ross

describe the genius of the designer, his freedom of creation.

Ashby and Heinz von Foerster), he assembled a rather unusual

They bear witness to the power of the mind that conceives the

network of Boolean automata.7 Kauffman's automata are logi-

project, a power that is all the more evident because it imposes

cal artifacts; the term "Boolean" refers to the functions the dif-

upon matter a way of being that is foreign to it. In contrast, the

ferent automata obey. Each of them "calculates," using one of

words typical, natural, express a mathematical requirement:

sixteen Boolean relations, an output value (o or 1) based on its

order should be "generic." Consequently, the relation of ends

input values. The fact that they are networked means that each

and means becomes misleading. Not all performances are of

of them, in synchronization, will (if it outputs 1), or will not

equal value. Anyone who wishes to understand the obligations

(if it outputs o), send a signal to those automata with whom its

ARTIFICE AND

LIFE

"output" is connected, based on the signals it has received from


other automata in the previous step. Until then, the performance
of networks of Boolean automata had been predetermined. But
the young Kauffman connected a hundred automata "randomly"
and found that the collective behavior of the resulting network
was one of unexpected simplicity, given the ensemble of possible a priori "states." Moreover, this behavior was robust: up
to a certain point, it resisted changes in its connections until
it "shifted" into another, different behavior (the landscape of
states is characterized by "attractor basins").
Kauffman's model was the origin for the held of "neoconnectionism," an explosion of new technological tools and mathematical theories that allowed researchers to "understand"
what had initially been discovered. Along with the "cellular
automata" for which Conway's Game of Life was the prototype,
it served to bolster the belief that "artificial life" was not mere
rhetoric. It ushered in a new model of the artifact that satisfied,
as is frequently remarked, a bottom-up rather than a top-down
8

approach. The artifacts creator no longer needs to be represented as a designer endowed with twiddling fingers that enable
him to carry out his project, to impose downward what he has
conceived topward. The creator "profits" from a new form of
causality we can call "coupling causality," which is neither linear nor circular as in cybernetics. It is the fact of coupling that is
important, not the type of interaction (physical, chemical, logical, electronic) or the purpose for which they are arranged. The
creator is interested in behavior that is already qualified, alreadyendowed, with a relatively robust landscape of possibles "emerging" from that coupling.
If the generic properties exhibited by the Boolean network
make it a "promising factish," "God's heart" should singularize
the new interest these properties arouse on the part of someone who addresses a "randomly connected" network, the new,
practical relationship between the artifact and its maker. For,

ARTrFrCE AND

L r FE

257

the "neoconnectionist maker" is not only looking to map the


stable behaviors of the network. He wants to modify, to model
those behaviors in such a way that on the map of possibles, "bottom" assumes the meaning for "top." The most typical example
of such a relationship is the one in which the network acts as an
"agent" for the recognition of shapes.
The example of shape recognition is interesting in that it
refers to an apparently simple performancesomething we do
without even thinkingbut which had always been difficult for
artificial intelligence to get right. For example, what is a "B"?
Yes, it is possible to formulate criteria for identifying the shape
"B." But those criteria must satisfy a rather formidable requirement, they must allow for the recognition of an indefinite multiplicity of Bs, one more "poorly written" than the other, some
of which resemble "D," others resembling "8," and others even
resembling "A." This is why it is crucial that neoconnectionist behavior be robust. The fact that the relationship between
an initial distribution of the values, o or 1, of automata and the
resulting stable behavior is resistant to modification of the initial configuration "promises" that if that relationship could be
constructed as a "recognition" of the configuration in question,
that recognition would be indulgent by definition, robust with
respect to variations. No longer is it a question of the production
of criteria that make explicit how the recognized shape is to be
specified but of the "learning process" that will make the difference between a welcome indulgence and one that is unwanted. It
is a question of establishing an optimal coincidence between the
attractor basin for all initial configurations leading to the same
behavior and the ensemble of all initial configurations that,
for us, are "Bs." In this case, learning involves a modification
(based on a process that is fundamentally random but automatically controlled) of the connections or weighting of connections
among automata until the network adopts the same behavior for
everything that we recognize as "B," and adopts other behaviors

252 A R T I F I C E

AND

LIFE

for everything that is not "B" for us.


A "random" network can learn, but it's important to understand that it doesn't learn all alone and, of course, it has no
knowledge of what it learns. The learning process involves two
elements and cannot be reduced either to a design, no matter
how tentative or negotiated, or to spontaneous evolution, no
matter how controlled. The maker proposes but the network dis poses, in the sense that, given the maker's propositionthe initial configuration that was imposedthe network evolves toward
a form of stable behavior that belongs only to itself, which the
maker may acknowledge but about which he harbors no ambition of predicting. For the maker, such behavior, regardless of
what it is, will be the answer, the translation, emerging from the
networked ensemble, of what was initially proposed, and it is
based on that response that learning will begin. For all the initial
configurations that the maker judges or wants to be similar, the
translation must remain the same, and for other propositions,
which he judges or wants to be different, the translation must
be different. No matter how approximately we write them, we
recognize that twenty-six distinct letters compose our words.
The network must be able to distinguish them. Leaving aside the
technical aspects of the algorithms used to modify the network
so "learning" can take place, the important point to remember
is that we are dealing with an interaction in the strong sense.
"The network is capable of learning!" " It is an artificial neuronal network, the first appearance of the absent body of artificial intelligence, that has just been invented." Such statements
are not the laborious conclusions of specialists, but they clarify
the premises of their interest, the conviction these networks
brought about almost immediately. Namely, that the network's
operation is a vector of meaning and, yet, incapable of justifying the meaning that "emerges" from that operation, creates
the topology of a "body." The "internal coupling" whose robustness can be used to make the transition from the ensemble of

ARTIFICE

AND

LIFE

53

interactions to the meaning of that ensemble "for" its operation is distinct from any relationship to a milieu, for there is no
"milieu"; a completely artificial environment determines the
initial configuration of the network. The invention of learning
practices creates a "body" by exposing what I have called the
causality of couplingthe causality that singularizes the networkto another, heterogeneous "causality" that couples the
network to operators that will set about teaching it to actualize
their own objectives.
With the appearance of the neoconnectionist artifact, every
speculative argument concerning our mysterious ability to
"recognize" things without being able to specify the criteria of
resemblance, which has engaged philosophers from Plato to
Wittgenstein, has been captured. There is no need to have an
"idea" of a table to be able to state "It's a table." The recognized
object "emerges" as a collective response, in the here and now,
without a model or a localizable memory. More specifically,
here "self-organization" causes a "quasi object" to emerge for a
"quasi subject," which should not be confused with the network
as such. The network itself is inseparable from the "quasi purpose" it fulfills, but the meaning of that "quasi purpose" relates
to the one for whom emergence occurs.
It is this "emergence" of a body through coupling between
the network and the maker that can, I believe, give to the term
used by Kauffman, "God's heart," an interesting interpretation,
even if it's not the one he intended. Whatever he intended, he
used a charged analogy that contrasts the heart not so much to
the fingers but to the rational mind of the designer who causes
his fingers to move on the basis of his project. Such analogies always reveal much more than what their user may have
intended. Judith Schlanger, in her marvelous Penser la bouche
pleine, used the example of an Egyptologist who "demarcates"
his object, the Egypt of Egyptologists, by disqualifying all the
other "frctive" Egypts. 9 Nevertheless, all of them, she claims, are

252

ARTIFICE

AND

ARTIFICE

LIFE

includedthe Egypt of myths, the Egypt of films and novels, the


Egypt of dreams. They all coexist alongside the demarcated Egypt
that disqualifies them within a dense milieu that makes interesting the demarcation that apparently excludes them. And it is
this density, this muffled and stubborn "cultural memory," that
allows us to understand the interesting innovation. The demarcation, if it were to create a vacuum, would be stable, attached
to its evidence. It isn't, because whenever it produces new consequences, these are liable to create resonance in the dense
milieu that feeds it, a milieu that, for the speaker as well as for
the hearer, becomes the vibrating matter of a new actualization.
In our case, the "heart," in contrast to a "reason" capable of
accounting for its operations, does indeed mobilize a dense cultural memory, in which the capacity that identifies reason continues to hesitate between recognition of unique legitimacy and
indictment for arrogant pretense that bars access to a different
order of truth. But it is not as one of philosophy's "great themes,"
first with Pascal and then in psychotherapy, that the problem of
the heart finds the means to insist. 10 On the contrary, what is
innovative is the way in which the problem is liable to be reorganized around one of its components. What Kauffman's "God's
heart" expresses is that the consequences "promised" by the
factish concern the way in which the "psychosocial" identity of
the makers of artifacts will be demarcated.
Andrew Pickering has compared the development of a new,
classical detector, one that uses a physical or chemical process
to identify an entity or process that is also chemical or physical,
to a kind of two-step dance." The scientist adjusts the machine,
then withdraws and allows it to operate. He observes what it
"does," in this case, what it detects, and interprets the reasons
for what he judges to be its defects. He then goes back to work
and readjusts the machine, and continues to do so until the
machine detects what it is supposed to detect. There is certainly
an interaction, but once the machine is stabilized, the scientist

AND

LIFE

53

has learned a great deal and can tell the story quite differently.
Now, the machine assumes a passive role, the action is entirely
redefined in terms of what the scientist "did not know" at the
start, problems he hadn't noticed, distinctions he hadn't thought
to make. The maker, when involved in the "dance," may indeed
have experienced extraordinary things, becoming a detector,
confronting a world a distinctive feature of which he seeks to
capture. But his psychosocial identity incorporates the way in
which the story will achieve its conclusion, with the final separation between himself, on one side, the world and the machine,
also physicochemical, on the other. The world bears witness
through the machine, the machine's operation is explained by
the world. The same is not true in the case of neoconnectionist networks, however. Here, the site of the "dance" is a coupled
causality that will never be disentangled. The maker will never
know how his device operates. And the device doesn't detect in
the ordinary sense. Its purpose is not to become the witness of
distinctions that could be said to belong to the world and need
only be recognized. It must produce conventional distinctions,
those to which the maker attributes value, among the resolutely
confusing shapes proposed by "the world." The culmination of
the process is not the separation of the maker, on one side, the
machine/world, on the other, but the maker/machine, on one
side, whose values are mutually adjusted, and the world, on the
other, always as confusing and bound to remain so. In fact, success for the maker occurs when "his" machine has succeeded in
recognizing the "B" that he had so carefully mangled when writing! And when the network fmally, spontaneously "recognizes"
what has been put before it, its operation can never be compared to a fragment of "nature" that may well have been selected,
staged, and purified, but should still obey the same "reasons" as
nature. The maker's judgments have passed into the machine,
the only "reason" for its operation being the conventions it has
"learned" to obey.1*

ARTIFICE

252

AND

LIFE

ARTIFICE

AND

LIFE

53

The learning network is not a hybrid comparable to the clock,

device ("this just gave me an idea . . ."). In this case, construc-

for example. It does not necessitate a historical and constructiv-

tion should be told in the form of a stoiy: a story in which the

ist reading that struggles against the triumphal syntax wherein

demarcation between the maker and the machine continuously

we distinguish the laws of mechanics, on the one hand, from the

transforms itself; a stoiy in which the maker's identitywhat

design associated with the human project, on the other. It is as a

he seeks, the possibles he intends to actualizewould "emerge"

hybrid, exhibiting the processes of stabilization and negotiation

along with the behaviors of his device; a stoiy in which the roles

from which it originated, that the device is presented. And the

would remain radically asymmetrical but would no longer put

maker is someone who has caused it to emerge as a hybrid, to

before us the owner of a project and the device that is supposed

the extent that a part of himself has "passed" into the machine

to realize it. A constructivist stoiy.

and has bound with the properties of genericness inherent in

Identifying the possible creation of a new psychosocial type

the machine to form a composite that no one is supposed to ever

of maker is a form of speculation. But the possibility of such a

be able to separate. What God has united . . .

speculative stance is part of the resonance effects resulting,

When scientists talk about God, they are often talking about

within the dense cultural milieu that entangles the themes of

themselves. The God of Einstein, a mathematician, occupies

fabrication, autonomy, emergence, and the link between cre-

the site Einstein hoped to construct. The demon-god of Laplace

ator and creature, from the redistribution of agencies that may

knows the world as Laplace, the astronomer, thought he was

be associated with a new type of artifact. In our stoiy, the cre-

capable of knowing the planetary system (which he believed

ation of the clock that ideally satisfies, autonomously and solely

to be stable). Maxwell's demon sorts particles that the physical

on the basis of the laws of mechanics, the intentions of the

chemist cannot at the macroscopic level. Langton's God plays

clockmaker, has had effects of which we are the heirs. Theology

on the world's keyboard. Kauffman's God has a heart, which

has been able to emancipate itself only by turning God into an

refers, I believe, to the interaction and hybrid world of recipro-

absent God. Biology is still an heir, and has given natural selec-

cal capture that is productive of meaning. A world in which the

tion the figure of the clockmaker, or, more accurately, according

"factish" made promising by its generic properties explains

to Richard Dawkins's expression, the blind watchmaker, adjust-

nothing as such, but implies and assumes a maker who interacts

ing, permutating, modifying the mechanisms of a population

and evaluates, and whose values are "passed on" to the world,

of "clocks" that, in the most highly diverse ways, tell the only

becoming, in the strong sense, an integral part of that world,

time that "counts" for the clockmaker, the rate of transmission

inseparable from it, an ingredient of an order that nonetheless

of genes over succeeding generations. And it is to biology that

remains "typical," "generic," and, as such, impenetrable, "even

Stuart Kauffman turned in attempting to read in it the conse-

to God."
Learning the alphabet is a poor example, however, because
the maker's values cannot be affected by the process. It is not
impossible that the new psychosocial maker of these new artifacts will one day refer to open-ended learning, where the
maker's "values" would be partly generated by the answers of his

quences of a possible "marriage between self-organization and


selection."' 3 For Kauffman, the blind watchmaker must " m a n y "
the generic properties of coupled causalities. Through its metaphors, language acknowledges the dense milieu in which such
references are distinguished: let no man put asunder.
Yet it is also, whenever living organisms are involved, at the

252

ARTIFICE

AND

LIFE

ARTIFICE

AND

L I F E 53

point where the clockmaker goes blind, where the figure of God

is compared to observable biological data that, when compared

and the maker must both disappear, that the question arises of

to the behavior of the toy, become interesting, capable of pro-

determining how the "factish" ofa coupling causality can become

viding, in certain situations, information in the language of the

an ingredient for the problem of emergence and evolution.

model.

For Stuart Kauffman, the issue is that of "theoretical biol-

The common feature of all of Kaufmann's toy models is that

ogy," but the notion of theory is profoundly ambiguous in this

they accept the hypotheses of Darwinian evolution but, contrary

case. If it were to function as it does in the theoretical-experi-

to neo-Darwinism, do not assume that selection is all-powerful.

mental sciences, it would imply the construction of a power to

They present the effects of a hypothetical selective pressure on

judge that should minimize the reliance on history and turn the

the exploration of a landscape of possibles, where the very point

terrain into a theater of proof, much like a laboratory. But rede-

is that everything is precisely not possible because exploration,

fined in terms of the practices of negotiation I associate with

from mutation to mutation, has as its subject the transforma-

the problem of emergence, it can signify an approach to what

tions to which beings characterized by internal coupling (for

biological evolution requires as we are able to puzzle it out. In

example, "interconnected genomic networks") are suscepti-

this case, the mutation imposed on the notion of "theoiy" by the

ble.' 5 It is the network itself rather than any given trait that is

theoretical biology with which Kauffman nourishes his dream

characterized by a coefficient of "aptitude," and the character-

would imply a mutation of the "theorist." Whether this mutation

istic connection rate of the network measures the number of

is clandestine or mutilated, whether mutant theoretical prac-

genes on which the meaning (in terms of aptitude) of a mutation

tice is persuaded to claim it resembles what it disagrees with,

affecting a gene depends. In other words, Kauffman's models are

the way Darwinian practice was persuaded to claim it retains the

not based on any new biological hypothesis. They are limited to

power to judge, or whether it has the freedom to assert itself, is

taking seriously what every biologist knows: the correspondence

an issue for the ecology of practices.

between a trait (more or less adapted) and a gene is in no way

The research that, for Kauffman, ushers in the new field he


calls "theoretical" is collected in his massive The Origins of Order,
which can be considered the leading work of contemporary
"theoretical biology.' 4 But the book will be unreadable for anyone who expects theoiy to provide the miracle of an approach
thai comprehends diversity within the luminous affirmation
of a principle to which it is subject. The book contains a series

representative of the living organism. Whereas the neo "Darwinian evolutionaiy biologist generally tends to minimize the
complications resulting from this minor problem, Kauffman's
models propose making it "the problem," primarily by studying
the effects of selective pressure as relative to the type of being to
which it applies.
A single general hypothesis finally falls out of Kauffman's

of studies of formal situations, which introduce relationships

exploration and it is upon this hypothesis that his desire for a

judged to be typical of biology, but in a highly simplified man-

theory is concentrated. If selection favors the ability to differ-

ner, through the use of toy models (here toy signifies both that

entiate, if it "encourages" the network to explore a spectrum of

the model is something to play with, rather than one that claims

diversified "activities," selective pressure should cause emerg-

to provide a faithful representation, and that we can play, that it

ing behaviorsand, therefore, the connection rate character-

can be manipulated). The behavior "emerging" from the model

izing the coupling from which they emerge as wellto evolve

252

ARTIFICE

AND

LIFE

toward the edge of chaos. Here, perfect order is behavior that

ARTIFICE

AND

LIFE

53

artifice to converge. Broad statements such as "natural selection

is completely predictable and robust. Kauffman refers to it as

must have . . ." are replaced by the indeterminacy of "we do not

"frozen." The system is locked into one and only one mode of

know a priori." We do not know how to formulate the question

operation. In contrast, perfect chaos is compared to a fluctu-

of "value" in general terms, in that it may refer to a particular

ating, erratic liquid, in which any alteration of an element can

trait or to generic properties of interconnected ensembles, such

trigger a cascade of consequences throughout the network.

as those that characterize the "edge of chaos." And, in this last

When order dominates, the freeze percolates throughout the

case, we do not know which coupling situation is the subject: the

network, but it leaves behind isolated, unfrozen, pools. In the

genomic network, specific ontogenesis, the dynamic of inter-

predominantly chaotic regime, however, it is the liquid regions,

specific coevolution? This is what needs to be conceptualized.

fluctuating chaotically, that percolate, leaving frozen islands

But the "promising factishes" of Kauffman's toy models are

here and there. The edge of chaos thus corresponds to a generic

vulnerable, as is self-organization far from equilibrium, to the

behavior that preserves the "best" of both worlds: the possibility

theoretical ambition that refers to itself, now and always, as the

of cascading innovation and relatively stable modes of operation

power to economize the terrain. This vulnerability is primarily

resistant to chance.'

"If it proves true that selection tunes genomic systems to the

expressed by the possibility of grand considerations that appear


to communicate scientific practice and wisdom. And in this case,

edge of chaos, then evolution is persistently exploring networks

it is a "stoic" wisdom that celebrates a universe that "awaits" us,

constrained to this fascinating ensemble of dynamical sys-

in the sense that we are the expression of chance, yes, but also

tems."' 7 In other words, selective pressure does not confer dif-

an expression of the generic order promoted by theoiy, a fra-

ferentiated "adaptive" values only on those beings that emerge

ternal universe because coupling is everywhere, but a dangerous

from coupling, but also on the coupling itself, as requisites for

one because of cascading consequences. "Our smallest moves

an evolution capable of providing its fecundity to the "marriage

may trigger small or vast changes in the world we make and

between self-organization and selection."


Kauffman's "toy models" obviously do not constitute a theoiy
in the sense that the multiplicity of forms of marriage might find
their respective contracts referred to a single institution that
would define the truth of marriage as such aside from any anecdotal differences. Quite the contrary, it is the apparent generality
of "selective pressure" as a vector of evolution, the possibility of
assigning to it the responsibility of evolution independently of
what it bears upon in each case, that is annulled, whereas a generality of a different kind is offered in its stead: a hypothesis like
that concerning systems "balanced" at the edge of chaos can ori-

remake together. Trilobites have come and gone; Tyrannosaurus


has come and gone. Each tried; each strode uphill; each did its
evolutionary best. Consider that 99.9 percent of all species have
come and gone. Be careful. Your own best footstep may unleash
the very cascade that carries you away, and neither you nor anyone else can predict which grain will unleash the tiny or the cataclysmic alteration. Be careful, but keep on walking; you have no
choice. Be as wise as you can, yet have the wisdom to admit your
global ignorance. We all do the best we can, only to bring forth
the conditions of our ultimate extinction, making way for new
forms of life and ways to be."' 8

ent questions, not answer them. It can bring into existence, as a

We could say that, in this case, Kauffman, as he did when

problem, the emergence of "adaptive values" that cause life and

he spoke of "God's heart," thinks with his mouth full. But there

268

ARTIFICE

AND

LIFE

is a difference, and it is crucial for the ecology of practices. In


the latter case, what is being expressed is what, for Kauffman,
understanding life demands. In the present case, stoic wisdom
includes an ensemble in which eveiythingfrompaleontological

17

data to the historical, technological, and political dynamics that


"identify" usbears witness in one way alone, that of an allegory of exploration exposed to selective pressure (we all do the

The Art of Models

best we can) and the price of that exploration (the unforeseeable


catastrophe). Here too, questioning the obligations of a practice of emergence entails questioning the kind of appetite this
practice induces for the "terrain." Is the factish's "promise" the
submission of the terrain to a theoretical-ethical-speculative
generalization or does it create an appetite for the terrain, where
the indeterminate promise to which it gives meaning might be
actualized.

It would be a misunderstanding to confuse an appetite for the


terrain with the creation of "good" science, respectful of beings
and participating in the secret harmonies of Being. If the practices that bring about the terrain-as-problem evoke a precedent, it is not one of Utopian reconciliation, where knowledge
would break any connection to power. Rather, it is the problem
of another form of power, analogous to the kind of power that,
according to Francois Jullien, Imperial Chinese civilization
favors, as evidenced by the omnipresence of the word chi.1
Chi is a word with as many meanings as our term "energy."
It refers to a dynamic configuration associated with nature as
well as with art and calligraphy, the composition of poetry, government, and warfare. The use of the word in Chinese thought
contradicts any possibility of contrasting phusis and techne,
spontaneity and manipulation, submission and action, conformity and efficiency, whether these refer to human government or the grand cosmic design. Chi implies the disposition of
things, of characters, of intrigue, of political or military power
relationships. And it refers equally to the arrangement that
produces their respective propensities and to the intervention that will, without force, noise, or, apparently, effort, take

269

274-

T H E A R T OF

T H E A R T OF M O D E L S

MODELS

advantage of this arrangement and lead the situation, as if by its


own dynamics, to the desired issue. A part of chi, therefore, is
the art of relying on chi for some advantage, the art of manipulation and enticement. The art of the great warrior is letting his
enemies kill one another or betray their agreement while he
remains invisible, so that the enemy army grows demoralized to
the extent that the final battle is no more than a formality. Here,
reason does not triumph over force, it weds force, it becomes
force, and does not respond to any criteria other than those of
efficient manipulation.
The art of chi despises violence, not because it would contradict a moral ideal but because it is not effective, because it
indicates failure by opposing the propensity of things rather
than confirming that propensity by taking advantage of it. Nor
is it eager to discover a truth beyond dispositions and mechanisms, or seek confrontation or harrowing dilemmas. But it
would be especially stupid, because this art escapes our excesses
and closes the perspectives in whose name we have committed
great crimes, to see in it the position of wisdom we are said to
have betrayed. On the other hand, it is worthwhile pointing out
that the practical mutation that could transform the dual identity of the artifact and its maker, as well as the question of the
"marriage" between biological selection and self-organization,
find their most apt metaphors in the art of conforming to the
propensity of things.
No doubt the Chinese would have understood Kauffman's
statement that "Evolution is not just chance caught on the wing.1
It is not just a tinkering of the ad hoc, of bricolage, of contraption. It is emergent order honored and honed by selection." 8
But they would have certainly understood it without the slightest sentimentality. "Honoring" and "honing" have nothing to
do here with moral respect; it is a question of using another's
force to bend him to our own purposes. This may be characterized as (and all such characterizations are pejorative for

us) "manipulation,"

"suggestion," "seduction,"

271

"appropria-

tion," "instrumentalization." The interesting point is that we


are accustomed to using these pejorative terms whenever they
refer to relations between human beings. However, they are now
presented as metaphors for a new type of relationship between
phusis and techne. The psychosocial image of the technician has,
until now, emphasized a practice conceived as submitting an
ideally inert material to a purely human project. And it established the figure of free choice and will as the problematic point
of contrast between the "emergence" of assemblages that were
respectively natural and human. The "technician of c/ii" has not
renounced his will in order to make room for the democratic
or revolutionary Utopia of a "self-organized" nature that produces order, beauty, and truth through the free spontaneity of
its self-creation. He is "without principles," no longer respects
the master word used to organize the hierarchy between knowledge and application: "Understand the principles nature obeys
in order to bend her to our purposes." It is enough that he can
make nature bend, follow her folds, marry them so he will be
able to create others.
It is interesting to approach from this point of view the mutation the term "universal" underwent within the problematic of
"self-organization." The law of gravity is said to be universal in
the sense that any mass, no matter where it is in the universe, is
supposed to obey it. However, the "promising factishes" of the
physical chemistry of systems far from equilibrium and network dynamics also allow one to speak of universality. 3 The very
beautiful word attractor accurately expresses what this notion of
the universal entails, the type of necessity with which it communicates. This necessity is always relative to a mathematical
or logical model, a hypothetical schema of relations expressed
by the model. Furthermore, when we deal with situations that
make evolutionary sense, the model aims less at representing the situation than at relating it to a problem. The universal

274-

T H E A R T OF

MODELS

defined by the model cannot claim to be that to which the situation is subject. It only claims to be relevant for an understanding of that situation. Although the model introduces a robust
attractor, characterized by generic properties that apply regardless of the circumstances, it designates a situation one of whose
ingredients may have been the question of the universal that has,
literally, captured it, infected it with these generic properties.
Various situations may be "judged" according to the terms of
the universal into whose grasp they have fallen. However, they
are not necessarily defined by the categories of this judgment
because they are capable, in return, of defining it in their own
terms.
The problem of emergence may be approached through
the art of models. The identification of a universal is no more
the answer to this problem than a propensity is an answer for
the "technician of chi." Such a universal is characterized by
the insistence of a question for which an answer may eventually emerge. The necessity with which it communicates implies
that, if the model is relevant, the modeled situation, in one way or
another, must have taken it into account and assigned a meaning
to it. Does this situation express it immediately? Do the generic
properties serve as an opportunity? Has an activity regime
acquired its meaning and purpose because of them? Has it succeeded in becoming a requisite for other activity regimes for
which the model would then provide an ingredient? Or does an
aspect of the situation that the model failed to take into account
become interesting and intelligible precisely because it allows
the situation to avoid being captured by the universal? The universal is a question, a proposition. As for the intelligibility being
constructed, it is related to the way in which the situation has
disposed of that proposition. The necessityif the model is relevantarises from the fact that, in one way or another, determining "how," the way in which the proposition has been disposed
of, must have taken place.

T H E A R T OF

MODELS

273

At this point, the model severs its connections with the theoretical-experimental practices that have made it a weaker substitute for theory, a representation that is not supposed to resist
the challenges that a theory must overcome. A model, as it functions in the theoretical-experimental sciences, has a domain of
validity that is carefully delimited, for, through its definitions, it
employs simplifying expedients whose scope is explicitly relative
to this domain. On the other hand, anyone who speaks of "theory" assumes the risk of claiming that the theory must remain
a reliable guide, even when used outside the practical domain
for which it was constructed. Once it is a question of the "field
sciences," however, the model is no longer defined in contrast
with a theory. The model is no longer defined by its simplifications or by ad hoc hypotheses. It no longer belongs to a practice
designed to "prove," because the validity of a given proof would,
in any event, be valid only for a given situation. Rather, it is a
question of producing a problematic tension between what the
model requires and what the field discloses. By identifying its
requisites, a model makes a wager and assumes a risk: what it
requires of reality should be necessary and sufficient for making
intelligible what has been learned in the field.
We can compare this use of the model with what Gould
defines as "Darwinian discovery." "We define evolution, using
Darwin's phrase, as 'descent with modification' from prior living
t h i n g s . . . . We have made this discovery by recognizing what can
be answered and what must be left alone."4 Darwinian evolution
requires the prior existence of living things. All of the reasoning
it employs presupposes this. It gambles, therefore, that biological evolution, in putting forth its own problem, has no need of
a solution to the question of the origin of life. In other words,
it positively denies a hypothesis like that of "vital force," which
would be simultaneously responsible for life's origin and its history.5 What has been "discovered," in the sense that the model
actively implies the reality it proposes, is the possibility of using

T H E A R T OF

274-

MODELS

a "disconnect," the possibility of separating the question of life's

T H E A R T OF

MODELS

275

the histoiy that has given life its "origin") it does not, or no lon-

origins from what happens once living things exist. The model

ger, intervenes in the terms of the problem. However, the ques-

of evolution cannot investigate the origins of life, for it requires

tion of determining how the problem will be formulated is part

selection, which assumes the presence of living beings; it

of what the model must explore.

requires the specific relationship that every living thing invents

I want to turn now to models that specifically concern the

with its milieu, its congeners, most often its predators and, in

problem of emergence. Unlike a model that might be called

some cases, its prey. 6

"scenographic," because it tests the consistency between the

Whenever it's a question of evolutionary models associ-

histoiy it can be used to predict and the history of the "field,"

ated with the field sciences, realist ambitionwhat the model

whose terms, witnesses, and indices are identifiable within the

requires of reality and the obligations entailed by the model's

modeled situation, the model of emergence attempts to articu-

claims to relevancerelies on requisites, on what the model takes

late a hypothetical emergence with requisites that are associ-

the risk of treating as securely given in order to proceed. This

ated with other practices, that is, requisites whose meaning is

ambition is not trivial. Most models in the social sciences and

initially relatively indeterminate with respect to the question of

economy fail to satisfy this requirement. Equations are written

emergence for which they are, hypothetically, a possibly neces-

expressing the consequences of rules, norms, laws, or conven-

sary but always insufficient condition.

tions which, the model claims, "explain" the evolution of social

It is here that we again encounter the question of "uni-

or economic situations. But these rules, norms, laws, and con-

versals" associated, primarily, with self-organization. These

ventions vaiy over time, and the model would only make sense

universals are part of a strategy that relates emergence with req-

if this variation were noticeably slower than the evolution the

uisites. They are relative to the construction of the model from

model is supposed to explain. Which, in general, is not the case.

the perspective of mathematical practice: the model in question

If the time scales are comparable, the model is worthless. This

belongs to a class characterized by a generic property, a "prom-

was Norbert Wiener's objection to the hope of Margaret Mead

ising" property in that it is impossible to "escape" it other than

and Gregory Bateson, who urged him to focus on the social and

by radically transforming the model. Once recognized, a univer-

economic sciences and make them fully scientific disciplines

sal of this type creates a terrain for the question of emergence,

that would finally contribute to solving the urgent problems fac-

for it defines one of the issues that "must have" polarized the

ing society.7

situation. If the model is relevant, if its requisites are legitimate,

To overcome Wiener's objection, a model must assert the

what emerges had to have "confronted the problem" and been

risks associated with it, the power relationship that character-

determined by determining the meaning that would be attrib-

izes the situation if the model is to be relevant. Only the situa-

uted to it. The universal gives the situation the significance of a

tion can authorize the modeler to separate what the model will

critique.

define as variables and constants, or forget certain aspects of

But the role of mathematics in the question of emergence

the situation in order to highlight others. The dimension of the

doesn't end there. It can also "shift" the issues associated with

situation that is responsible for the satisfaction of the model's

a scenographic model toward a problematic of emergence. I will

requisites can be forgotten to the extent that (as is the case for

give three distinct examples of such shifts, three typical cases of

274- T H E A R T OF

MODELS

T H E A R T OF M O D E L S

277

what singularizes the questions of emergence when compared

introduced by those equations, modeling has allowed sceno-

to theoretical-experimental questions: in each case, as the

graphic practice to "rise" to the "how" of emergence. This rise

modeler learns to formulate a problem, she discovers that this

is expressed by the correlative appearance, for the modeler, of a

problem has been (partly) formulated before.

quasi subject, the populations of competing predators, respond-

Take the problem of eco - ethological models that make use of

ing to a quasi object that is none other than the very object of her

a predator and its prey. The initial scenographic model, designed

modeling: the "universal" problem of interspecies competition

to account for situations where statistical series are found to

for predator populations.

exist because humans, as predators, have been interested in

Therefore, the modeler should not trust her model, not

the frequency of capture over long periods of time, is the so-

because the model might be wrong or irrelevant but because she

called Lotka-Volterra model, which makes use of predator-prey

does not know, a priori, how it is relevant. The Lotka-Volterra

interactions. The model typically results in a form of periodic

model apparently designates an "object," but it must be used

behavior. Predators eat abundantly and reproduce easily, but

with tact in order to expose the possibility that a "quasi subject"

at the expense of their prey, whose numbers decline. Conse-

might have appropriated the problem corresponding to the

quently, hunger and famine occur and the number of predators

model. The question of knowing "how to describe" is no longer

decreases, which benefits their prey, whose numbers increase.

one that concerns the scientist alone. Correlatively, the nature

This allows the predator population to increase again, and so on.

and scope of "objective" definition are transformed. Objectiv-

This first example, however, is simply a starting point toward the

ity is beside the point. Interspecies competition is a problem for

general case that introduces competition among predators. We

specific groups, but it does not allow a solution to be deduced;

can then ask about the evolution of populations coupled by their

it raises the question of finding out how, with what ingredients,

shared dependence on a set of resources. However, the empiri-

using what expedients, a solution has "emerged."

cal relevance of the model of interspecies competition encoun-

Tact is a quality most often exercised among humans, but it

ters limits that have nothingto do with the complicated details of

points to a much more general problemthat of a relationship

such coupling. In fact, field studies lead to a change in the nature

created with a being for whom a problem is assumed to exist,

of the model. Rather than being a scenographic model of cou-

a problem that can be identified, or so it is believed, although

pling to which competing populations are subject, it becomes a

how the problem presents itself to this being is unknown. Tact,

description of the coupling that some species manage to escape.

therefore, expresses an obligation that limits the power of who-

Seasonal changes in reproduction, the choice of resources, the

ever is situated by her knowledge of the other's problem. She

amount of food needed at different times of the yearall these

"knows," accepts, and desires a relationship that incorporates

"details," which the model "smoothed," can become interest-

the open question of the "how" and "tactfully" respects the fact

ing to the extent that they counteract the effects of interspecies

that time is needed for the answer to this question to "emerge"

competition. The relevance of the model changes. It is no lon-

for the concerned being. Teachers who lack tact do not feel this

ger tied to coordinating its predictions with empirical data but

obligation, and most often those who are tactful fail to cap-

to identifying specific behaviors that falsify those predictions. K

ture the identity of the "how" that has been invented during

Moving from the question of solving equations to the problem

the course of the relationship. The goal of the modelerand it

274-

T H E A R T OF

T H E A R T OF

MODELS

MODELS

279

is in this sense that her tact is part of a scientific practiceis to

1 he modeler to perform can become the starting point for a new

define the way in which the situation she models answers the

question that uniquely designates the behavior being explained

model's question. "Tact" then comes to imply a transformation

as the specific behavior of a living being.

of requirements and obligations compared to those that govern

The "system" of nine (nonlinear) equations taken as such

experimental procedure. We could even say that it is no longer

defines a literally "unmanageable" system that may gener-

the scientist alone who imposes requirements. Of course, the

ate extremely diverse behaviors, even though it is supposed to

scientist must require that what she addresses has a stable exis-

explain "what the amoeba is capable of," "what it does," that is

tence in terms of the relationship that is established. Wherever

to say, behavior that is stable and reproducible. Consequently,

the conditions of a field science are found, the features stud-

the practice of the modeler cannot be reduced to one of simple

ied must be robust with respect to the type of intervention that

confrontation between the model's predictions and described

allows them to be studied. But the field also allows itself to be

behavior. The modeler doesn't require that the amoebas verify

characterized in terms of its own requirements. The relevance

her equations, she is obligated by the amoebas to recognize

of the scientist's problem depends on the fact that this prob-

that not all the possibles defined by the equations are valid for

lem has actually required, long before the model that makes it

them, that some are excluded and others privileged. The amoe-

explicit, an answer that gives it meaning. Correlatively, the field

bas, therefore, obligate the modeler to pose the problem of

"obligates" the scientist to recognize its "preexistence," to rec-

her model, for it is now a question of understanding how they

ognize that she, the scientist, will only encounter it by acknowl-

themselves, in one way or another, "manage" the diversity that


the equations define as unmanageable. Can the modeler reduce

edging that preexistence.


This same quality of tact is at the center of my second example: biochemical modeling. Take the behavior of the amoeba
Dictyostelium discoideum in the presence of cyclical AMP. Cyclical AMP, a creature of biochemistry laboratories, intervenes in
the intracellular behavior of amoebas and in their intraspecies
relationships. The rhythmic production of cyclical AMP in the
milieu serves as a "signal" for the population, that is, it modi10

the number of equations, distinguish, for example, which are


"slow" and can be decoupled from the others? In this case, she
will have to "trace back," through the values of the parameters
that must be selected in order to support the appropriate behavior, to what the model now allows her to identify: an ensemble of
biochemical "quasi choices," which have intervened in the very
invention of Dictyostelium discoideum.

fies the intracellular behavior of "receptor" amoebas. The data

The modeler's practice, the detailed negotiation with the

of biochemical analysis culminate finally in a "scenographic"

parameter values, the calculation of their consequences, in

model of nine interconnected equations with nine variables.

a sense closely follows the problem of selective evolution as

Can the model be used to explain the behavior of the amoeba in

it is made explicit by the model. Selective evolution then cor-

terms of the molecular interactions it introduces? In one sense,

responds to a figure closely allied to tact. The model builder's

fortunately, yes, as this behavior is not that of the amoeba itself

initial equations form the matrix of a "luxuriance" of possible

but a partial description, one that has already been worked and

temporal behaviors and imply that a mutation that modifies a

reworked to allow the question to be asked. But the interesting

reaction rate, or introduces, eliminates, or alters a coupling, may-

point is that the work the successful explanation has obligated

have uncontrollable, and usually catastrophic, consequences for

274- T H E A R T OF

MODELS

T H E A R T OF

MODELS

281

the amoeba. Selection no longer has much to do with the figure

an optimum means understanding the efficiency, not deducing

of the watchmaker, blind or not. The selective histoiy of the bio-

it. The emergence of collective behavior has to be "followed"

chemical mechanism of the amoeba's behavior has much greater

according to the way in which interactions among ants "modu-

need of the precautionary prudence of an apprentice pickpocket

late" (but do not determine) individual behavior. And the col-

working on a mannequin covered with bells. Tact, the clever

lective behavior that "emerges" from such interactions turns out

negotiation to obtain one thing rather than another, more often

lo be remarkably efficient indeed, capable of preferring a large

one thing rather than the unavoidable other, correlates the prob -

source of food over others, or systematically exploring a milieu,

lems the modeler faces when confronting her equations and the

similar to a projector revolving around a nest. The erratic and

problems selective histoiy (from which the role conferred on

nonrobotic (programmed) behavior of the individual ant

cyclical AMP by the amoebas emerged) had to resolve.

becomes, in this type of model, an essential component of group

The practice of modeling in biology is often the work of

efficiency. It requires that the individual be somewhat "entre-

researchers who take inspiration from economic models, but the

preneurial" in order for the group to "explore" the opportuni-

problem with economy is that it radically lacks tact. Its appetite

ties in its milieu. But a more general concept also arises, which

for theorems, used primarily to determine optimal conditions,

changes the stakes when studying collective behavior. Not only

takes the place of relevance. Why not have the model hypoth-

does a given population of ants in a given environment select

esize, for instance, that unemployed workers "disappear" from

the food sources that "count," but the way in which it selects

the market if that is a condition for a theorem?" The economist

them suggests a hypothetical "tracing back" to the problem of

requires, with a unilateral brutality that is the opposite of tact,

that multitude of species we call "ants." The interactions among

that the modeled situation give her the right to publish a theo-

ants are such that a small quantitative change in a parameter

rem. When employed in biology, this lack of tact immediately

(which may correspond to a random genetic variation) quali-

conspires with the omnipotence that neo-Darwinian theorizers

tatively transforms the way in which the method of seeking out

give to selection. What emerges must optimally satisfy a given

and selecting resources operates. "Ants" in the generic, multi-

adaptive value, and the existence of the optimum allows evolu-

species, sense could then coincide with the invention of a rela-

tion to proceed from theorem to theorem. On the other hand,

tion between individuals and the group, which is the "matrix for

"modeling the held" can, as we shall see, enable us to counteract

significant variants." The relationships that allow the transition

the theorem-based inspiration of the economist and "return" to

from the individual to the group would not only belong to a spe-

the problem that singularizes a behavioral trait.

cies, they would (partially) identify that species with a "choice"

Take the typical behavior of ants in search of food. 12 The


uniqueness of this behavior is its intelligibility on the group
level. Although individual behavior may appear somewhat
erratic, the behavior of a group of ants is a key example of efficiency, and it seems to deserve an explanation maximizing some
adaptive value. If we assume an optimum, we can always construct it, but if we don't, other questions arise. Not to assume

made on the basis of a genetic matrix of "hypotheses," subject to


selection in each different environment, a genuine "machine"
that is no longer adapted but adaptive. Here too, the question of
emergence appears with the acceptance of the problem, with the
possibility not of "reducing" one level to another, but of introducinga quasi-practice of inter-"level" articulation.' 3
The

three

examples

aboveinterspecies

competition,

274-

T H E A R T OF

MODELS

T H E A R T OF M O D E L S 282

amoebas, and antsapply to different aspects of biology, but

In other words, it is not at all certain that a practice whose ideal

they have one thing in common: they describe how we read a

is the convergence between the requisites of the model and the

way of functioning that is stable and capable, to a certain extent,

requisites of the body "itself" retains its relevance in the case

of reproducing itself from generation to generation. The mod-

where it addresses a living being whose experience includes

eler wagers that since it is robust, it must be understood as hav-

the feeling that it "has a body." Whenever it is a question of

ing invented the means to be robust, and the relevance of her

the human body, in particular, and its marvelous or terrifying

activity as a modeler depends on this wager. If we look at the way

ability to allow itself to be "modeled" by cultural practices, the

Deleuze and Guattari define the concept of a body, by relating it to

question of determining what a model wishing to address "the

"informational coordinates of separate, unconnected systems,"

body" should address becomes critical. To speak of "modeling"

we can say that the wager is that the situation is "embodied," and

cultural practices is itself significant. The "model," in the sense

as such defines the emergence of a disjuncture between internal

in which it refers to a scientific practice, can no longer be dis-

and external variables in relation to the milieu, which has noth-

sociated from other "modeling" practices. The human body is

ing to do with the distinction between internal variables and

always that of a being belonging to a given family, a given group,

the limit conditions of physical-chemical systems.141 To define a

a given culture, and this belonging also implies the way in which

system by its limit conditions does not imply tact, and the prin-

the body is "fabricated," the way in which it is "understood," and

ciple of exploration to which this definition corresponds is one

how the requisites of its "normality" are identified. And at this

of variation (to increase pressure, temperature, the intensity of

point the power relationship "within the modeled situation,"

the temperature gradient, or the imposed relationship of chem-

which the scientific practice of modeling requires and benefits

ical concentrations). To define an "organism" does not imply

from, disappears.' 5 Trap, temptation, and curse, the question

tact either, if the organism refers to a body judged in terms of

that arises is less one of the disappearance of this power rela-

a relation between ends and means, where eveiy organ fulfills

tionship than of the derisive ease with which it is obtained. The

a function through the harmonious division of responsibilities

human "collaborates" with the project of elucidating what it

and tasks. Addressing a "body" imposes this specific art I have

requires and, in some cases, even what it is subject to. We are

called "tact." The model must explore the disjunction as such,

in the process of preparing to explore the transition to the limit,

approach it from two sides at once. It must negotiate the relevant

where the relationship between construction and definition will

internal variables with respect to observable external behavior,

again change its nature.

but also approach that external behavior from the point of view
of the milieu it defines for by itself and for itself, that is, identify the selection and values of the variables it requires from the
milieu in which it emerged "as a body," identify how, from its
point of view, all milieus are not equal.
The body, in the customaiy sense, is certainly composed of a
multitude of bodies in the sense I have introduced above. But it
is not at all certain that it functions as a body in the same sense.

TRANSITION

TO T H E

LIMIT

does not constitute a given problem, imposed by a change in


the properties to be interpreted. More specifically, if proper-

18

Transition to the Limit

ties do change, the way such change is characterized involves


a critical question about the veiy issue of characterization. To
make this question and the commitment it demands perceptible is to make perceptible the "critical point" at the limit.
Here, therefore, the existence of the limit belongs to the "present" of whoever effects, but first experiences, the transition to
the limit. It defines this present relative to the perplexity, the
"perplication," of the questions and distinctions that the limit
has stripped of their tranquil differentiations. a The critical
questioning of knowledge does not have the generality of critical thought, which always silently assumes the ability to judge on

In physics, approaching a limit imposes a number of precautions whenever several variables simultaneously tend toward
the infinite or toward zero at the limit. To avoid any confusion,
these variables must be individually managed. Physicists must
take the risk of emphasizing a single variable in order to construct reasons why the description of the problem that gave
them their meaning (for example, what is a gas?) loses its relevance at the critical point, even though they know that they
are changing their meaning collectively. Similarly, I must try to
"slow down" the loss of relevance to which the transition to the
limit corresponds. In this case, that means trying to remain for
as long as possible within the framework of my initial question,
that of scientific practices in which the scientist can risk requiring, so as to identify where and why this requirement changes
meaning. But this is only a first step, for the question of the limit
returns. What the initial question assumed was a pathway to the
limit, but not the pathway, the one that would coincide with the
general definition of that limit. On the contrary, for it is from
that limit that one can attempt to turn the pathway itself into a
problem.'
Unlike the situations studied in physics, the "limit" here
284

behalf of what is not questioned. It is part of the risks the present obligates us to take.
In Cosmopolitics, Book I, "The Science Wars," I described the
problem that, for me, requires a transition to the limit, namely,
the "modernist" practices 1 took the responsibility of characterizing as constitutively polemical. For, in order to present
themselves as scientific, they need to disqualify the opinions,
the beliefs, of others, the nonmodern practices of which some
claim to serve as rational substitutes. Identification of the problem and the question it raises situate me because they express
the conviction I have tried to implement until now with respect
to other scientific practices. The way in which those other scientific practices create their questions and their risks satisfies
requirements and obligations whose singularity instantiates a
difference with what precedes or surrounds them, a difference
that has no need to be reinforced through polemics and disqualification. The same is not true of "modernist" practices, whose
claims postulate that the one who asks questions, because she
is a scientist, which is to say rational, which is to say modern,
escapes the illusions, traditions, and cultural assumptions that,
on the contraiy, define those she is dealing with. Modernity here

a86

TRANSITION

TO T H E

LIMIT

TRANSITION

TO T H E

LIMIT

287

is an integral part of the definition of science in the sense that

an external device that is liable to literally "pass into" the human

it gives the right to invoke a stable difference, a difference that

and, therefore, seems to promise a stable definition in the face

allows one to judge and claim kinship with the power relation-

of life's contingencies. But the example is too complicated to

ship whose invention the experimental laboratoiy celebrates.

slow us down reliably. The number of young humans who will

The critical questioning I associate with the transition to the

never be "one" with the alphabet even though they are supposed

limit refers to this commitment to dissociate modern science

to have "learned to read" is too high not to suspect that the ques-

and modernist science. The critical point signals the appearance

tion of "emergence," here apprenticeship, cannot eliminate

of modernist practices, discussed in "The Curse of Tolerance"

everything that "knowing how to read" allows us to ignore. Pos-

(Book VII), where I clarify the "cosmopolitical" question that

sibly, among the ingredients of apprenticeship are the multiple

gives its name to this series of books. The domains I'll address

components that, in another form, belong to the art and experi-

are those where the definition of a scientific practice can no

ence of reading (about which "knowing how to read" is equally

longer benefit from a stable difference between the scientist's

silent), namely, those that enter into the effective encounter

practice and what she interrogates. And it is the heteroclite

with a particular text. The "true" reader is one who may well

ensemble of practices, modern or otherwise, and the beings,

be able to read "in general," but for whom the encounter with a

factishes and fetishes, to which they refer and which are ingre-

text has nothing general about it. The slowdown turns out to be

dients of their existence, whose modes of coexistence will then

impossible, for even assuming that what "emerges" could really

be (begin to be) examined. But before risking this approach, in

be modeled, what this descriptive, scenographic model would

which perplexity would have to construct the practical obliga-

benefit from, that is, the irrepressible nature of "knowing how

tions that satisfy the perplication of questions and distinctions,

to read," turns out to be an obstacle to the possibility of tracing

we must slow down, examine situations where the problems

the description back to the question of emergence. The model

that will trigger the transition to the limit are already present,

would reproduce the final emergence but provide no clue for

but where modeling is not yet a caricature and still has a chance

the many questions that cluster around what "learning to read"

to express what the being the model describes requires of the

requires.

milieu with which it has been coinvented.


It would be worthwhile to take as an example those studies

On the other hand, there is another episode, one that is truly


generic in human life. It is the one that leads infants, in one way

in experimental psychology that attempt to penetrate the mys-

or another, to transition from the mode of existence of a young

teiy of an activity such as reading, which has some interesting

mammal, not fundamentally different based on appearances

characteristics: the laborious manner in which it is learned,

from a newborn primate, to that of a young human engaged in

the way it breaks down under the effect of neurophysiological

language learning and the relationships their specific identity

disturbances (recognizing letters but not words, words but not

presupposes but that must "be produced" for each of us indi-

sentences), and the fact that the reader, once she has "emerged,"

vidually. This episode is so fascinating that it has brought about

"reads the way she breathes," that is, cannot prevent herself from

the equally fascinating but relatively indecipherable series of

identifying a word but, on the contraiy, must make an effort to

experiments attempting to get primates to "talk." And it is the

identify individual letters. This is a veiy interesting example of

subject of an indefinite number of speculations and variants,

T R A N S I T I O N TO T H E

a86

LIMIT

where science, myth, and religion freely intersect. But most

T R A N S I T I O N TO T H E

LIMIT

289

categories. The psychoanalysts' infant would have to "fantasize,"

extraordinaiy is that, in the face of such divergent interests,

it would have to experience the original illusion of a fusion,

one way or another small children continue to successfully

and the adult, barring nearly irreparable damage, would have

manage this transformation, in any event, the vast majority of

to renounce that initial experience of well-being. 4 It seems to

them.

me that what Stern is suggesting is a new kind of model, which

It seems, then, that we are dealing with an extremely robust

introduces the requisites of the infant but also assigns a cru-

"histoiy" whose success is tied to the very invention that defines

cial role to the unique nature of its interaction with adults. For,

what it is to be human, a histoiy made to be repeated and that

according to Stern, the manner in which adults "respond" to the

could, in this sense and to this extent, be defined as an extrauter-

infant's "behavior" poses the same question as apprenticeship

ine extension of human ontogenesis. It is as if the infant had

itself, which involves both "repetition" and "acculturation." It

its own requisites, as if it were capable of a power relationship

would incorporate both their cultural, familial, and personal

with an environment that, barring any dramatic circumstances,

interpretation of what those behaviors signify and what those

enables it to learn and become. And yet, we also know that, at

same behaviors lead them to feel and do irrepressibly, that is to

the same time, another kind of histoiy is beginning, inseparable

say, robustly. Correlatively, "emergence" would occur through

from the first. In fact, when the infant manages to stand and take

asymmetrical capture over time. Through its behavior, the child

its first steps, and even earlier, it is indeed possible that this

suggests a response from adults, who in turn suggest to the child

histoiy has already begun. But in the case of learning to speak, I

a new way of being, and the process repeats.

feel I can take it for granted that the situation is clear: the infant

In this case, we can speak of emergence as a productive and

does not learn to speak in general. Together with words, there is

functional misunderstanding, whose terms change continuously

an indefinite ensemble, implicit and explicit, of ways of being,

but irreducibly entangle human genericness and cultural-

of entering into relationships, of interpreting and anticipating,

familial specificities, producing a child who has become capable

that is created or stabilized. The two-year-old child is no longer

of experiencing itself and others as endowed with continuity,

a small, generic being; it is the child of a family, a culture, a tra-

historical materiality, and intentions, but who experiences them

dition. It would seem, then, that the requisites of the newborn

in a way that integrates fundamentally heterogeneous ingredi-

do not communicate solely with the notion of a necessary but

ents, rhythms and refrains in adult value judgments, whether

insufficient condition but with that of a necessary and necessar-

implicit or explicit, concerning affects, legitimate or illegiti-

ily insufficient condition. Which is to say that they incorporate

mate, expressible or inexpressible. These ingredients can arise

in their veiy definition ingredients that must be determined by

from a mode of action that may or may not be deliberate, and

what is no longer a "milieu."


In any event, this is the reading Daniel Stern proposes in The
Interpersonal World of the Infant.3 Psychoanalysts quickly recognized that Stern's book made use of a disquieting approach, one
likely to classify as "professional legend" the version of the myth
of paradise lost and original sin that were the basis of their own

they may be consistent or contradictory among themselves.


They coexist in distinct ways, each of them understood in what
Felix Guattari recognized, in his own terminology, as "incorporeal universes" and "multiple, dislocated, and entangled existential territories." 5
Several paths are possible with such a model. One that

a86

TRANSITION

TO T H E

TRANSITION

LIMIT

should obviously be avoided is using the model in the predictable operation of normalizing the description and confusing the
"successful" relationship between parents and infants in our
own culture with what the human infant requires in general.
Another path leads to a consideration of the relationship
between apprenticeship and misunderstanding.

Misunder-

standing is a loaded word, but here it has no Freudian-Lacanian connotation implying the impossible fulfillment of desire,
or the always failed relationship, or the painful lack at the core
of any illusion of belonging. This type of dramatization is very
interesting from the professional point of view of the psychoanalyst, who effects a decentering and creates a highly specific
power relationship that stabilizes the therapeutic process in a
unique and radically unilateral way. 6 1 will attempt to follow how
the concept of a model changes meaning without the operation
suddenly having dramatic or disparaging consequences. Stern's
description "models" the young human, but here there can be
no question of condemning a given form of alienation but of
approaching practices that "introduce" a human into a world it
can inhabit only if it learns to comply with the requirements of
what it will encounter there. After all, even, and especially, in a

TO T H E

LIMIT

291

and constructing. It is only "after" that the words retroactively


appear sufficient to define the knowledge that is transmitted
through them. 7 Between the "after," where the teacher cannot
but dwell, regardless of her good intentions, and the "before,"
where those she addresses are positioned, transmission implies
a genuine practice of misunderstanding." We can even speak of
a "categorical" misunderstanding to the extent that, contrary
to other kinds of learning (walking, riding a bike, driving a car,
juggling, mountain climbing), mathematics is unique in that it
confronts the one who learns it with explicit formulations that
comprise both conventional rules and normative injunctions.
However, as in other kinds of learning, it is a matter of "embodiment," of rules and injunctions "passing into" the body. Whereas
the set of definitions and rules appears to introduce a purely
"spiritual" operation, the pure product of abstract formulation,
reasoning, and proof, what must be produced when one "gets it"
are ways of perceiving and being affected in a functional, nearly
automatic, way. To be able to recognize " ( a - b i n an algebraic
text and automatically adopt the mental gestures and practices
appropriate to the problem expresses the success of the corresponding modeling operation.

highly formalized science like mathematics, it is through mis-

In such a case, misunderstanding is not another way of

understanding that definitions and rules are held to be self-suf-

expressing the question of human existence, the failure of lan-

ficient, operating in such a way that compliance, understanding,

guage, which never lets us say what we "desire," or the tension

and application go strictly hand in hand.

between the never satisfied quest for truth and the risk of cynical

Mathematics, which in Greek meant "that which is readily transmissible," in this sense constitutes a very interesting
example of a "misunderstanding." Even when a mathematical
definition is transmitted for the billionth time, what we call
"comprehension" remains an event, the production of a "before"
and an "after." It is only "after," once we have understood, that the
normative words through which this knowledge is transmitted
assume their effective meaning, which transforms them into references, instruments, and constraints for exploring, reasoning,

abandonment. It does not designate the kind of staging that confers the power to recognize sameness throughout each step of a
psychoanalysis or the "phenomenology of the spirit ."This kind of
misunderstanding could, however, communicate with the concept of transduction created by Gilbert Simondon in L'Individu et
sagenesephysico-biologique.9

Transduction does not refer to the

human, to language, or the search for genuine rapport, but to the


problem of individuation, through which an individual characterized by discreet relationships with its milieu is produced. In

a86 T R A N S I T I O N

TO T H E

LIMIT

fact, Simondon used the physical-chemical phase transition


and the concept of a critical point as an experimental field for
creating his concept. But anyone who might claim to draw from
transduction the power to recognize that the production of an
infant with an individuated relation to language responds to the
"same" problem as the genesis of a crystal would be misusing
the concept. Comparing the crystal with the infant has meaning
only because the first step in the process of transduction is not to
define the process of individuation but to learn to resist the way

TRANSITION

TO T H E

LIMIT

293

scales of reality that are able to communicate with one another


through transduction the power to obscure the process of communication, that is, the power to explain the individual. 12 In this
sense, the first difference between the human and the crystal is
that genetics and macrostructure are rival explanations, while
the beauty of the perfect crystal relies on its ability to effect a
harmonious convergence of two rivals: the forces of interaction
between atoms and the energy equilibrium between the crystal
and its environment.

in which the problem has generally been presented. Transduc-

Wherever it is relevant, transduction attempts to bring about

tion applies both to the crystal and to the human to the extent

a form of thought that is capable of resisting the temptation to

that neither the terms that enable us to explain the individuated

choose between rival principles of explanation, a temptation

crystal (interatomic forces, a configuration that corresponds to

Simondon qualifies as "hylomorphism": the Aristotelian duality

the minimum potential energy resulting from those forces) nor

between form and matter. For Simondon, this duality has served

the terms that can be used to explain the human (genetic pro-

as the matrix of every position that has been adopted since then.

gramming or social, cultural, economic, or symbolic structures)

Some of these base explanation on a "form" that imposes itself

allow us to describe the process of individuation.

on matter thought to be available, others on "matter" conceived

In all cases, what must be resisted is the temptation to explain


the genesis of the individual from previously individuated conditions, the way the mold would explain the statue or hypotheti10

cal statements a solved problem. Atoms, genes, and structures


make the individual the simple realization of the possible they
define, which is to say they miss the process of individuation.
Simondon also tried to provide a generic description of the
process of individuation through transduction. "This is," he
wrote, "the physical, biological, mental, and social operation by
which an activity is propagated gradually within a field, basing
that propagation on a structuring of the field enacted from place
to place." 11 The crucial point is that this operation always implies
communication, but first as a problematic tension, between

as being capable of causing form to emerge. Is it "symbolic


order" or the norms of mathematics that are transmitted unaltered that "inform" an available mind, or is it the "matter" of the
operation, a form of generic competence of the human psyche,
that is responsible for the possibility of learning? It is as a vector
of resistance that does not limit itself to celebrating the "failure" of these alternatives but creates a new appetite and riskier
obligations 13 that transduction might assist in constructing the
problem presented by the relationship between apprenticeship
and "misunderstanding." 14 The asymmetrical capture correlated
in time that, for Stern, "models" the infant would then be a primary example of the communication of two "scales of reality"
whose heterogeneity is their primordial given.

two scales of reality, one "greater" than the future individual,

This would be a good place to slow down, for it is not enough

the other "smaller." And it is this "primordial heterogeneity"

simply to have good intentions. Transduction, because it enables

that will be retranslated, once individuation takes place, into

us to simultaneously contemplate crystallization and human

two rival explanations, each of which confers upon one of the

modes of individuation and individualization, is speculative,

a86 T R A N S I T I O N

TO T H E

LIMIT

TRANSITION

TO T H E

LIMIT

95

and the intrusion of speculation is part of the transition to the

the latter cannot claim any superiority over the others by virtue

limit that I am attempting to initiate.' 5 There would be no more

of its conceptual reference. It should be added to the others,

unfortunate confusion than to treat this intrusion as a victory,

along with its own requirements and obligations. And its own

as the conquest of a point of view that considers the distinction

risk of failure. In other words, transductive thought provides

between construction for a scientific purpose and speculation

no benefits with respect to the strictly empirical problem found

to be pointless. Nothing could be worse than to view Simon-

in these examples. It provides no guarantee. Its role is to create

don's ideas as the basis for a scientific approach to "emergence."

words that might stabilize thought capable of resisting the slo-

The inherent challenge of speculative thought is the creation of

gans and legitimations through which the risks associated with

concepts that allow us to speak, simultaneously and at the same

a practice become rights (of reason, progress, objectivity) for

time, of what our habits oppose (for example, crystallization and

which the practitioner is merely the representative.

thought), but this creation is an experiment in which our hab-

Speculative reference to transduction thus puts at risk the

its are both ingredient and target. It does not seek the discovery

power of models that claim to authorize an economy of perplex-

of a point of view that would guarantee the right to unify what

ity. Experimental factishes can, through a constitutive vocation,

we oppose and to establish a judgment concerning the "proper"

claim to "explain" the world, and it is possible to assert that the

way to answer questions that produce hesitation, perplexity, or

world "explains itself" through them. Reference to transduc-

expectation.' 6 The practical effect that singularizes speculative

tion reminds us that, here, explanation, made possible by the

thought is to contradict the temptation of a judgment that rec-

coming into existence of each factish, primarily celebrates the

ognizes and anticipates. This thought straddles abysses, but the

primordial heterogeneity between the requirements of the sci-

"same" that it constructs, the "anticipation" it feeds must accept

entist and the world that is supposed to satisfy them. But the

the constraint of "accommodating" no one, of not confirming

reference to transduction can also help recognize and celebrate

any particular practical requisite, not justifying any power rela-

the occasions when the scientist, temporarily putting aside any

tionship. That is why this thought is fundamentally descriptive,

professional plausibility, searches for the words to express the

and the possibility of drawing normative consequences from it,

question that the experience of what she is involved in invinci-

regardless of the register of the norm, indicates either its failure

bly imposes.

or the (mis)appropriation of its use.' 7

So, when I tried to put into words the expression that

Transductive thought produces the effects proper to specu-

appeared on Kauffman's lips, "God's heart," I created the fig-

lative thought to the extent that we cannot, without possible con-

ure of an interaction involving a "Maker" whose values pass

tradiction, make use of it without also introducing at the same

into the world. I made use of a figure that expresses transduc-

time the "transductive" nature of its use as soon as it becomes

tion, which can be used reciprocally to assert that the maker's

a part of practice. Practices of apprenticeship may take their

"values" do not explain what is made, even though they "explain

inspiration from various forms of hylomorphic thought, with an

themselves" through the making process. But in doing so, I bor-

emphasis either on the "form" to be transmitted or on "matter,"

rowed the words that Stephen Jay Gould dared employ at the end

as is the case with constructivism, or from Simondon's critique

of an article in which he tore apart the "just so" stories of socio-

of hylomorphism; all are distinct examples of transduction, and

biology. The biological theory we need, he wrote, should replace

a86

TRANSITION

TO T H E

LIMIT

the questionable charm of such stories with the profound joy


arising from an understanding of evolution as integration: "the
world outside passing through a b o u n d a r y . . . into organic vitality within."' 8 Here, Gould doesn't claim to be a "vitalist," but he
effects, with a joy that accepts perplexity rather than the pretense that must deny it, a transition to the limit. Gould also uses
the word integration "with his mouth full." It is in terms of "integrative insight" that Gould, in the same book, evokes the way in
which Barbara McClintock allowed herself to be invaded by the
apparently disparate multiplicity of "data" produced by maize in
an attempt to understand that data. And Gould compares this
integration with the experience of Dorothy Sayers's detective
hero, Lord Peter Wimsey: "He no longer needed to reason about
it, or even to think about it. He knew it."' 9 And when Gould talks
about his own experience, he writes, "And so my work has been
integrative; that's what I'm best at doing. I do figure out Dorothy
Sayers's mysteries because Peter Wimsey is constructed as that
kind of thinker. If you read Whose Body?, her first novel, I'm sure
that Dorothy Sayers had a theoiy of thought and that she wrote
those novels to counter the Sherlock Holmes tradition that
thought was simply deductive and logical."

20

Here, Gould describes in the same terms, with his mouth


full, a theoiy of life that should help biologists do their work, a
theoiy of thinking, and equally his own experience as biologist
and writer when the outside, the scattered elements and bizarre
connections of a situation, move inside and contract into a living unity"he knows." It is here that speculative thought can
assume its "ecological" scope, bring into existence the perplexing joy of this convergence, and give it the means to produce
its own divergence, one that would prevent it from becoming
a pretense, a skeleton key that would open all doors and would
be confirmed in all cases: the birth of the kind of all-purpose
response produced by a transition to the limit that is brought
about unnoticed, accompanied by the exaltation that the feeling

TRANSITION

TO T H E

LIMIT

297

of truth provides. Convergence mustn't be avoided, it should


be celebrated, but in suspense, held in its problematic space,
"countereffectuated" and not precipitated into triumphal
solution. 21
The divergence to be recognized, and which indicates an
approach to the limit, affects all the terms required by scientific practice. Consider the term "confirm." Even Popper had to
admit that scientists are right to seek experimental confirmation when their theories are bold and fragile. Confirmation, he
proposed, is not a proof but the nourishment the fragile creature
requires. But here confirmation will always be experienced as
proof or as an argument that can be put forth to support a proposition. It will always express the power relationship between the
one who asks the question and the one who answers it. Adults
who encourage the young child to take its first steps also "confirm" its attempts, and this confirmation is itself likely to be as
vital for the child as "fact" is for the bold proposition. However,
it will never serve as a proof or an argument. What about the distinct and entangled modes of confirmation negotiated by teachers and family for children in school? What about the analysand
whose dreams "confirm" the interpretation of her analyst? What
about the experiences that, it is said, confirm "faith"? To follow
and map such divergences, it is necessaiy to countereffectuate
the proposed convergences and deliberately ascend the slope,
resist being carried down by the power of resemblance. 22
I have associated the art of modeling emergence with tact,
but tact is no longer a secure thread. Even when it is associated
with human relationships (doctor/patient, adult/adolescent), it
always refers to a power relationship that is able to control itself
and can create the space the other should come to fill in its own
way. The modeler, the doctor, the adult all propose and know
that it is up to the other to dispose. 23 Yet, it is the very meaning
of propositions that is affected by the transition to the limit, that
is, the meaning of the confirmation we expect from the other.

a86 T R A N S I T I O N

TO T H E

LIMIT

Suspending triumphal confirmationthrough the device or disposition implementing the propositionforces the history of
our satisfactions to ebb, and it is the "our" that begins to blink,
that causes what those satisfactions have identified to diverge.
We have benefited, and will continue to benefit, from all power
relationships, from every stratification that may allow for a stabilization of the difference between the question asked and the

TRANSITION

TO T H E

LIMIT

299

they think with their mouth full would probably be less interested in hollow generalizations, reflexive recursiveness, or
other irresolvable paradoxes. But the question doesn't end
there, doesn't affect only "us," our knowledge and its relationships. I'd like to return, one last time, to the starting point of
the transition to the limit I have attempted, the description that
Daniel Stern provided for the "emergence" of the infant.

answer that confirms it. We can find out how a "body" defines its

What is unique about this emergence is that it concerns a

milieu. But what it means to "live" or "die" does not follow from

becoming that is of interest to us all, and by all I mean all cul-

the thread of our definitions. Experiencing this marks the criti-

tures, all traditions, modern and nonmodern. Stern's descrip-

cal moment when constructivism escapes, as event, from the

tion may, initially, challenge hylomorphic models of all stripes

stories in which the practices that allow us to claim to know what

that modern researchers have proposed for giving the infant the

we know are constructed. The moment when the question of the

ability to establish their hypotheses. That is why, for example,

impersonal nature of the infinitive insists through "us," when

the Sternian baby challenges the Freudian baby, which is made

"knowing" begins to resonate with its opposites.

to establish that what will follow its emergence will confirm the

In "The Science Wars" (Book I), I limited the scope of an


ecology of modern practices to the question of determining
if new psychosocial types could be generated, new "we's" not
defined by polemics and hierarchies. Resolving the question
of the ecology of practices through the speculative becoming of
practitioners would be a trivial solution. Rather, the question
is one of asking which "type" of practitioner would not have a
phobic relationship"but if we introduce this type of problem,
we can no longer work"at the moment of reflux, when their
categories are confused. That is why it's important to acknowledge that speculation is not part of a fascinating "beyond" but
already inhabits those moments of confused joy when the scientist thinks with her mouth full. That is why the way physicists
have learned to define gas and liquid in terms of a transition to
the limit is also interesting. For this transition does not require
criticizing the gaseous state and the liquid state, but integrates
into their definition the question of the critical point at which
the distinction between those two states is, in fact, at issue.
Practitioners familiar with those "critical points" at which

power of "matter," the universality of the unconscious conflicts


of psychoanalysis, as well as the behaviorist baby, which celebrates the power of an exterior "form" to inform matter and the
availability of matter to form. It challenges the Lacanian baby as
well, which it prevents from dramatizing the misunderstanding,
the discordance between "interior" and "exterior." But, distinct
from the risk of its normative becoming, to which I've alluded
in passing, the Sternian baby presents another, more insidious
danger. It is capable of allowing us to claim that we have now
understood how "the others were not mistaken." Those others
are the "nonmoderns," who, for example, believe that the newborn is a stranger from another world, who speaks another language, a stranger whose identity must be discovered so it can be
named, and who must be welcomed and humanized. 84 Couldn't
we see in this a marvelous illustration of Stern's description, a
wonderful confirmation of the definition he proposes? Thanks
to Stern, we "now know" that the way in which we "welcome"
the newborn, the way in which we conceive of, anticipate, and
interpret its behavior, is a vital ingredient of its becoming. Isn't

300

TRANSITION

TO T H E

LIMIT

it wonderful that others have, without having read Stern, created the words and references that inhabit and guide parents in
this process?
But that is the danger, for this "wonder" is liable to be cel-

book vii

ebrated as follows: "Thanks for confirming the progress of our


knowledge, the validity of our new definitions. Thanks, and forgive
us for understanding you better than you understand yourselves, for
having constructed in your place the meaning of what you are doing.
In order to protect you, we will avoid telling you that we understood
what your beliefs really meant,' what they enacted without realizing
it. Your ancestors and your fetishes no longer surprise us, neither do
they disgust us. We have taken from them what we needed to leam,
and they confirm that our descriptions are right. They will serve as an
argument against our backward colleagues."
Suspending the confirmation, safeguarding the moment
when the impersonal"to speak to a child" or "to come into the
world"vibrates, are essential here. Not in order to avoid the
unavoidable, the feeling that "we have understood," but in order
to stand back and experience it in such a way that the suspension
of its confirmation is incorporated in its occurrence. For the triumphant confirmation I have presented above qualifies us. If we
yield to this triumph, we will trample, with the best intentions
in the world and with the additional satisfaction of remembering our own arrogance, the inappropriable space "where angels
fear to tread."

The Curse
of Tolerance

19

The Curse of Tolerance

Nothing is easier for modern man than tolerance. How could it


be otherwise? How could we not be tolerant? I am not referring
here to "others," to those in whom we encourage tolerance. I'm
speaking of "we," and this "we" does not refer to a concrete group
to which one may or may not belong, but to all recipients of the
message of modernity. It is a message that, as a "master word,"
is instantly applied as soon as we hear, understand, and accept
that "we" are not like others, those we define in terms of beliefs
we are proud, but possibly also pained, to no longer share. 1 Our
era is no longer one of crusades. The master word is occasionally paired with a nostalgic vibration. Tolerant is he, or she, who
measures how painfully we pay for the loss of the illusions, the
certitudes, we attribute to those who we think "believe." Therefore, happy are those whose confidence has remained intact.
They dwell where we, moderns, cannot return to other than as
caricatures, sects, and despots.
But nostalgia and tolerance toward others who are lucky
enough to "believe" barely hides our immense pride. We are
"adults," we are capable of confronting a world stripped of
its guarantees and enchantments. This tired refrain is well
known. As if by chance, it serves as a favorite theme among

33

THE

34

CURSE

OF

TOLERANCE

THE

C U R S E OF T O L E R A N C E

305

modern specialists of what Freud called the "three impossible

does not appear in the foreground, or it appears only in the ser-

professions": education (the transmission of knowledge), gov-

vice of the science that produces it. Conversely, "social workers,"

ernment, and healing. These three professions have in common

whenever they need to mediate, refer to knowledge they did not

the appropriation of practical concerns that all human societ-

produce but that is supposed to prepare them to deal with situa-

ies share. And the second thing they have in common, intro-

tions in which an individual, a family, or a group has a problem-

duced by Freud when he characterized them as "impossible"

atic relationship with its environment. They are not supposed

and assured his followers of his "sincere compassion," seems


to be, in our so-called modern societies, that their practitioners
consider themselves the privileged heirs of the loss of illusions
2

that defines modernity. As if their primary obligation were to


affirm the inevitable nature of the destruction their practices
inherited, as if their first requirement were to obtain recognition for and propagate the heroism of the impasses and aporias
their practices explore in detail and celebrate using modernist
slogans.

to "generate" knowledge themselves, but merely to produce


modifications in relationships, enabling negotiation whenever
there is a threat of confrontation or repression. As such, there
is no reason to criticize the distinction between knowledgeproducing practices and mediation-producing practices. Their
obligations are partly distinct. But only partly. To address such
distinctions, the question of what those obligations might have
in common should first be asked, a question I associate with the
question of tolerance.

To the extent that what Freud called a "profession" is dupli-

Throughout this book, I have approached scientific prac-

cated in our societies, the question of tolerance, which should

tices from two points of view: requirements that must be satis-

be "cursed," is twofold. It refers to a group of knowledge prac-

fied by whatever it is they are dealing with and the obligations

tices that are defined by their ambition to be recognized as "sci-

they acknowledge and that apply to the way they progress. Now,

entific" in the modern sense of the term. Here, I am referring to

I am confronted by practices that present the problem of tol-

what are generally known as the "social sciences," which cover

erance because, in one way or another, explicitly or implicitly,

a spectrum that ranges from psychoanalysis, psychology, and

they assume, between "we" and the "others," a difference in

medicine to various forms of sociology, teaching, anthropology,

kind as expressed by the possibility "we" claim of judging "oth-

and many other fields that serve as references to various forms

ers" in terms of beliefs without ever encountering them. From

of social assistance and intervention. But the term also refers

this point on, the leading question will be that of obligations,

to practices that, in one way or another, serve as caretakers for

because the constraint I have referred to as a "requirement" is,

our sociocultural "ecology," practices that address, for better

at this point, no longer a reliable guide.

or worse, the relationships that may be produced among het-

Until now, the practice that required was put at risk by

erogeneous groups. Naturally, the groups are intertwined, but

its requiring. The values that bring the experimental labora-

their intertwining can, depending on the situation, appear dif-

tory into existence are risk-based, and the obligations of the

ferently. Anthropology, for example, whether it wants to or not,

experimenter correspond to that risk. It is not just a question of

produces a relationship, but one that is most frequently funda-

finding out whether or not the account of the phenomenon pre-

mentally asymmetric. It provides "us" with a knowledge of other

sented, staged, in the laboratory confirms the arguments of the

groups, but the relationship from which that knowledge derives

one who produced it. It is a question of differentiating between

36

THE

CURSE

OF

TOLERANCE

THE

CURSE

OF T O L E R A N C E 305

two meanings of the term "artifact." In any case, an "experi

At this point, new types of "factishes" appeared, which I have

mental fact" is an artifact, a fact of human art, capable, unlike

referred to as "promising." These factishes are promising in the

facts in the customary sense, of authorizing an interpretation

sense that "based on" what they demonstrate, questions may be

and silencing rival interpretations. 3 But what succeeded in giv

asked that they are unable to answer as such. Learning if, how,

ing it this capability must resist the accusation of being respon

and in what formulation the promise has been kept must be

sible for it. The experimental account must not be an artifacl

produced "elsewhere." The answer must come from the encoun-

in the sense that the artifact now disqualifies the experiment I

ter with the histories, intrigues, reasons, and relations that the

claim, defines the experimenter's procedure as responsible for

scientists who interact, in their own field, with beings said to

an account that has been extorted, "fabricated," incapable of

be "emergent" learn to decipher and narrate. Requirements

supporting the challenges demonstrating that it can be reliably

and obligations then relate to the practice of articulation. This

attributed to the phenomenon being questioned. The obliga

practice requires, at its risk and peril, that what is deciphered

tions of the experimenter correspond to the definition of wha!

in the field can, at least partially, be defined as a response to

Bruno Latour called an "experimental factish."4. Certainly, the

the promise-question. How this response is expressed is inde-

fact is "fabricated," it can even mobilize an impressive crowd

terminateone must address the fieldbut that there will be a

of technical devices, each more sophisticated than the other,

response, in one form or another, that is, that the question will

but its fabrication is directed at the "invention-discovery" of

be relevant, is the risk that supplies a framework to the encoun-

a being that can claim to exist autonomously, independently

ter in the field. Correlatively, I have associated the obligations

of the practices that enable us to "prove" that existence. Pas-

this practice of encounter brings about with the values of "tact."

teur's microorganism, once it entered scientific existence, also

Whoever is endowed with "tact" knows, or thinks she knows,

became capable of claiming to have existed before humans and

what the other's problem is, but she also knows that this knowl-

to have been the vector of epidemics, even though humans read

edge will be worthless if it is delivered to the other. Therefore,

supernatural intentions into the scourge that assailed them.


With the question of emergence, the relation between
requirements and obligations is already complicated. 5 In order
to free emergence from the polemics and rivalry that have made

she also knows how to create the space in which the other will be
able to determine, in his own time and in his own way, how this
problem will be formulated and the meaning it will assume.
Between the "experimental factish" and the "promising

it one of the focal points of the "science wars," I have chosen to

factish" the relationship is pharmacological in nature in the

reexamine it through the question of its specific requirements,

sense that the pharmakon is unstablecure or poisonjust as

which are forgotten once answers are turned into military flags.

the sophist, physician or seducer, is unstable (see Cosmopolitics,

Determining if, and to what extent, "that which" emerges can

Book I, "The Science Wars"). As soon as the "promise" claims

be "explained" based on "that from which" emergence is pro-

to be endowed with any ability to determine how it should be

duced is not a matter for experimental observation or theory

kept, that is, as soon as the promising factish claims a power

but relates to a practice of articulation focused on the difference

and autonomy that assimilate it to the experimental factish, the

between "that which" and "that from which," that is, between

practice of articulation tips the values of tact toward those of

two distinct approaches corresponding to distinct practices.

proof. The field is no longer a site of problematic encounters but

38

THE

CURSE

OF

TOLERANCE

a theater of operations of conquest that are appropriately said to


be "reductive." Molecular biologists have succeeded in encountering bacteria in such a way that the promises expressed by the
DNA molecule were able to be kept by an articulation between
"genetic information" and the metabolic performance of bacteria. But this unique success wavers once "that which is true" for
bacteria is said to be true for all living organisms.
We now come to the problem of practices whose power of
judgment is said to depend on differentiating between the scientific, rational, or objective approach to a situation and the
beliefs, customs, habits, illusions, and so on that define the
actors in that situation. In saying that the question of requirements is no longer a reliable guide, I mean that the requirements
associated with both types of factish, experimental and promising, maintain a pharmacological relation with those practices.
The satisfaction of the requirements of a science is always associated with powerwe can require because we insist that what
we address has the power to confirm the legitimacy of our questions. But that which is addressed by both the "social sciences"
and the practices of intervention that are authorized by those
sciences push this "we can" into the indeterminable. For the
"power of science," the power of whoever it is who is said to "do

THE

CURSE

OF T O L E R A N C E 305

by the school, and so on. Rather, I am trying to limit myself to


the practice of producing knowledge. Not because I feel that
these problems can be generally separated but because they
must be distinguished if we are to avoid jumping to conclusions,
the amalgam that precipitates every problem at once. The power
of the experimental factish, which is capable of resisting challenges that question its autonomy, does not deny the multiplicity of the other ("technopolitical") powers that set it in motion.
But it makes it possible to create distinctions that avoid confusing, in the same accusation, Watson and Crick's joy at discovering the double helix of DNA and the mobilization (also overseen
by Watson) of biology and medicine around the challenge of
mapping the human genome. But questioning requirements
whose satisfaction would create an intrinsic bond between the
social sciences and power, identifying the pharmacological ease
with which terms such as "obey," "understand," "interpret,"
"predict," "control," and "verify" change their meaning when
they involve humans interested (in one way or another) in the
knowledge that can be produced about themselves, is not my
intention. It is only the starting point for extending into this
field the question of scientific practices, the values they bring
about, and the risks that identify them.

science," here becomes an ingredient that is inseparable from

Therefore, it is on the basis of the question of the specific

scientific practice. And this happens in two ways. It is on behalf

risk that identifies the social sciencesthe threatening possi-

of science that the scientist feels she has the right to ask others

bility that, even with the best intentions in the world, the effort

some veiy strange questions, to subject them to highly unusual

to create knowledge in their case might turn into the abuse of

situations, to describe them in a way that, in any other context,

powerthat I have chosen to construct the question of obliga-

would be judged highly uncivilized. And the power that precedes

tions that might singularize those practices. By acknowledging

and accompanies this is also an integral part of the attitude, the

the very specific link that unites "science" and "power" once

availability, the submission of those the scientist interrogates.

what it is that science addresses has to be defined as "sensitive

I am not referring here to the inseparability of powers, the


possible connivance, intentional or not, between the ethnologist and the colonizer, between the social worker and the police,
between the teacher and the work of social selection performed

to power," those obligations will bring into existence, as a value,


the ability to escape the trap represented by that link. 6
To give a sense of the tenor of these new obligations, I want
to offer a quote from William Blake:

io

THE

CURSE

OF

THE

TOLERANCE

He who mocks the infant's faith


Shall be mock'd in age and death.

CURSE

OF

TOLERANCE

The curse against "tolerance" is not based on a concern


for justice. With regard to those "others," it is not a question of

He who shall teach the child to doubt

defending them, as if the practices that we disqualify needed us

The rotting grave shall ne'er get out.

to do justice to them. It is a question, now and always, of our own

He who respects the infant's faith

practices, of obligations capable of stabilizing the irreducibly

Triumphs over hell and death.

pharmacological, unstable nature of the reference to the gen-

The words are those of a curse. And to escape the curse, it is

that is here liable to change into poison risks that are legitimate

eral advancement of science as a duty we must fulfill, a reference


not enough to tolerate "the infant's faith," for the "respect" that
triumphs over hell and death has nothing to do with abstinence,

whenever laboratoiy or field practices are involved.


The practices I have analyzed until now are also unstable, but

tact, or the protective kindness that always culminate in self-

I have tried to stabilize them using their own resources, their

betrayal, and betray as well the secret derision they conceal. "A

own passion, without creating obligations and requirements

robin red breast in a cage / Puts all heaven in a rage."H Cursed is

that differ from those they create for themselves and that create

he who frees the bird to please the rebellious child.

them. That is why I accepted the fact that what they addressed

Therefore, I have chosen to introduce the practical question

was, thereby, "put at the service of science," judged in terms of

presented by the scientific enterprise whenever it addresses

requirements whose satisfaction confirms practice. The choice

beings capable of wondering about the one who investigates


them "What does she want of me?" by treating tolerance as a
curse, and by tying the question of the obligations the social sciences might bring into existence and that might bring those sciences into existence to the risk of that curse.
The "curse of tolerance" evokes the highly nonneutral nature
of my questions. It is not a question of weighing the pros and
cons, but of bringing into existence a risk, that is, to treat as an
"abuse of power" what some social sciences may consider to be
success. A curse upon anyone who thinks they are free to redefine, in their own terms, the way in which the "other" inhabits
this world, even when they are willing to tolerate them, even
when they regret their own lost innocence. For innocence disqualifies the other, the one who does not yet know, the one who
has not yet endured the "great divide" that forced us to recognize
that a bird is not only a bird, and that the heavens are indifferent
to our constructions. 9

of the "curse of tolerance" as the operator of challenge indicates


that, from this point forward, being "put at the service of science" has changed meaning, has become something that might
serve as an obstacle to the invention of the sciences. However,
I don't want to waste my time "denouncing" practices that cannot resist this challenge. The production of obligations is a creative act. To create the curse of tolerance as an obligation doesn't
mean cursing whatever would betray this obligation. It means
constructing a question. How can escaping tolerance communicate with the obligation of what I have called a practice? Isn't
it, by definition, a unique adventure, a becoming that precisely
embodies an escape from the requirements and obligations of a
science? This is the specific question that defines this book and
will lead me to the heart of the speculative question of an ecology
of practices.

THE

C U R S E AS TEST

313

world is not pure being, but the sense which is revealed where
the paths of my various experiences intersect, and also where

20

my own and other people's intersect and engage each other like
gears. It is thus inseparable from subjectivity and intersubjec-

The Curse as Test

tivity, which find their unity when I either take up my past experiences in those of the present, or other people's in my own."'
If the phenomenological world "is not the bringing to explicit
expression of a pre-existing being, but the laying down of
being," everything must play out, now and forever, as if for the
first time, within an action-passion that is both dialog and infinite meditation. 2 There is no room here for any sort of "economy," obligation approaches infinitude through the twofold
task of "revealing the mystery of the world and of reason." 3 Phe-

To turn the curse of tolerance into an operator guiding the construction of what would pass its test is to gamble with what scientific practices may, or might, be capable of. It is a question of
prolonging them, rather than bringing them to a sudden stop,
which would be relatively easy. All that would be required to do
this would be to establish, between the ensemble of "objective"
scientific practices and the question of knowing what humankind can know about itself, the supposedly impassable barrier
that separates object from subject. But we mustn't fool ourselves.
Such a separation is extremely costly whenever it becomes
foundational, that is, when it is no longer limited to serving as
a reference for statements intended to "protect" the subject, but
is promoted as a starting point for "another kind of knowledge"
that would be addressed to the human being "as a subject." The

nomenology, Merleau-Ponty rightly concludes, "rests on itself,


or rather, provides its own foundation." 4 Which means, baldly
stated, that it can no longer be a question of practice in the sense
evoked in "The Science Wars," which I associated with the question of the psychosocial type. Anyone engaged in the phenomenological quest is not constrained by requirements that can be
satisfied and obligations that can be met. And they cannot be
a researcher, in the sense that the researcher always addresses
other researchers within the collective practice of research. The
problem is not resolved but suppressed, because the response
rests entirely on the inexpressible quality of someone who has
to be able to claim to "reveal" meaning, whereas everyone else is
constructing meanings. Escaping the curse of tolerance in this
way is precisely what I wish to avoid.

practitioner of this other kind of knowledge is asked to partici-

The curse of tolerance thus refers to the speculative possi-

pate in a form of asceticism, for it is a question of purifying the

bility of a psychosocial type of researcher, not the sublime figure

relationship to knowledge, of eliminating everything that, in the

of the sage, the phenomenologist, or the mystic. A speculative

practitioner as well as what she addresses, might become mate-

possibility does not simply fall from the sky of ideas. Speculation

rial for objectivation.

originates in unique situations, which exhibit the possibility of

This is true especially of phenomenology, whose ambition

an approach by the very fact that they have already undertaken

has been expressed by Merleau-Ponty: "The phenomenological

it. That is why, unlike the other books in Cosmopolitics, this one

312

314

THE

C U R S E AS

TEST

will engage me with other authorsprincipally Bruno Latour


and Tobie Nathanwithout whom it could not have been written. For, here, it is a question of "thinking with one's allies,"
not in the sense that, like Monsieur Jourdain, they would have
done unawares what I describe in their place explicitly, but in
the sense that what they have done, the risks they have accepted,
the issues they have made perceptible, have been obligations
for me. And they have resulted in the creation of the obligation I
refer to as the "curse of tolerance."
However, before introducing my allies, I would like to present a "quasi ally," an author who "could have" accompanied me
all the way but who, in fact, will enable me to evoke another connotation of "curse." As an exclamation, the word is often used
when we find ourselves trapped once again by what we thought
we had escaped. It is something that might be uttered by someone who thinks she has escaped from a carnival maze only to run
into a wall that bars the passage and sends her back to where
she came from. It is this dimension of the "curse of tolerance,"
the way in which we discover that we are again in a position to
be "tolerant," just when we thought we had overcome the chal-

THE

CURSE AS

TEST

organism. "Minute differences apart, much the same chemical


reaction occurs in both instances. In both cases the flesh reacts
to the acid chemically, b u t . . . the living organism 'knows'which
is a form of behaviorwhile the excised flesh does not 'know,' and,
therefore, does not behave. " 6 Similarly, for Devereux anxiety and
method raise the question of what the obligations of the behavioral sciences might be if their method didn't protect them by
"a similar downgrading of the observed. What sound behavioral
science calls for is not an (actually or fictionally) decerebrated
rat, but a recerebrated behavioral scientist." 7 A wide-reaching
program, you might say, with little exaggeration, for the "cortex"
in this case does not refer to an intelligence that behavioral specialists would be deprived of. The problem is not one of the distressing inferiority that singularizes the practitioners of those
sciences. It is on the basis of the anxiety to which their practice
exposes those practitioners that the double mutilation against
which Devereux struggled has to be read: that of the animal or
the human endowed with "behavior," that is, capable of observing the observer, and the questioner.
The problem raised by Devereux intersects my own, which

lenge, that I want to explore in the work of Georges Devereux,

is that of a definition of science that entails "mutilating" what

the founder of a field designed to help the modern researcher

it takes to be its object, even if it proclaims the necessity of

encounter the nonmodern "other"ethnopsychiatiy.


Devereux titled one of his most important books FromAnxiety
to Method in the Behavioral Sciences. He used the theme of anxiety
to address the "pharmacological" question of the methods that
what he globally referred to as the "behavioral sciences" bor5

rowed from the other sciences. For Devereux, anxiety is not the
near-universal affect known to Freudians and method is not the
repression of something that cannot be satisfied. For Devereux,
anxiety and method raise the question of what is required of the
specialist of the behavioral sciences when her method demands,
implicitly or explicitly, that she forget the difference between
pouring a drop of acid on a lump of dead flesh or on a living

respecting it after all. And in both cases, what is problematized


is less a moral question than a change in the nature of practices
that produce "facts" and "proofs" whenever they are directed
at beings that are not indifferent to the way they are treated. In
those scientific practices where requirements might serve as a
reliable guide, the fact that those requirements provide a stable
reference for the scientist with respect to what she questions
would express the stable difference between the two meanings
of the term "test." The scientist subjects what she questions to
various tests in order to verify that what she is dealing with is
capable of yielding a proof, but those tests have nothing to do
with the tests undergone by beings able to experience them

i8

THE

C U R S E AS

TEST

THE

C U R S E A S T E S T 39

as "tests." It is precisely when we are dealing with the specific

what will become a phenomenon, with observable attributes,

problem presented by what Devereux calls "behavior" that this

and what will be defined as an instrument, whose definition

"nothing to do with" changes meaning and becomes a defense

expresses the practical question of the observer, it corresponds

against anxiety. The possibility that scientists operate without

to the creation of data in the sense that such data have nothing

their own cortex arises specifically whenever they maintain the

to do with raw facts. The creation of data constitutes a twofold

postulate that there is a difference in kind between their tests,

actualizationfor the one who asks the question and for what-

which are meant to provide proof, and a testing experience

ever the question is aboutand both entail risk. The expres-

inflicted on powerless beings. This denial entails what Devereux

sion "And this is what I perceive" is in itself the fulfillment of

characterizes as a mutilation of thought.


Devereux's work provides a fecund example of the approach
I have tried to introduce with the concept of an ecology of prac-

an operation through which "this" and "I" are distinguished,


thereby creating a boundary that, for Devereux, is nothing other
than the "ego."8

tices. In fact, his work is the product of a practitioner trained

The ego defined by Devereux has no attribute independent

in physics who was shocked by the poverty of what are known

of the way in which the demarcation is produced. When the

as the "social sciences," including ethology. But unlike many

demarcation is practical, transmitted by practitioners, stabi-

others, Devereux avoided a two-part trap. He refused to rea-

lized by the "method" and the instruments of that practice, this

son in hierarchical terms; that is, he refused to believe (as do

ego presents the question of what I have referred to as the prac-

many physicists who turn to biology or the social sciences) that

titioner's "psychosocial type," defined in terms of requirements

a good dose of "real science" would save those poor, retrograde

and obligations. But this psychosocial type is here defined in

sciences. And he deliberately turned his back on the tempta-

a way that will allow Devereux to emphasize the uniqueness of

tion to respond to the question of behavior with an effusion of

the behavioral sciences, and what makes practitioners so read-

good feeling that would ensure that the scientist remembers

ily choose to "deprive" themselves of their cortex. For Devereux,

to respect the "subject" before her and interact with it while

the behavioral sciences are unique in that they expose them-

respecting a freedom that too easily excludes the rat suffering in

selves fully to the risks that characterize any practice where

the name of science.

measurement is involved. Theoretical variables always codeter-

For Devereux, all knowledge practices share a common

mine instrument and object. But in the case of the drop of acid

trait: in one way or another, they culminate in a cognitive utter-

poured not on a lump of flesh but on a living organism, those

ance: "And this is what I perceive." But this common trait is

variables should make fully explicit and put at risk the theoiy

not a reductive instrument, which could be used to establish

of the observer about what the organism in questionrat or

resemblance. It is a common thread when formulating relative

humanendures: not only what it is made to suffer but the fact

differences and practical contrasts among the sciences. The

that it "knows" that it is made to suffer. And when the behavioral

choice of this common thread relates to quantum mechanics,

sciences concern humans, if we are to avoid the self-destruction

which Devereux was quite familiar with, and which constituted

of the theoiy they make explicit, those variables must be capable

a decisive step on his path toward a constructionist approach to


knowledge. If measurement implies a "demarcation" between

of explaining the behavior of the observer herself. 9


To characterize that observer, Devereux once again invokes

i8

THE

C U R S E AS

TEST

THE

CURSE AS

TEST

39

quantum mechanics. The intervention that creates data and

iheir own countertransference, of the anxiety that has led them

explanation is always a perturbation of the observed, the demar-

to demarcate themselves in one way rather than another, is the

cation between what observes and what is observed also being

only practice that does allow them to speak of analytic "data."

the demarcation between what perturbs and what is perturbed.

Psychoanalysis in Devereux's rather particular sense pro-

This is well expressed by the concept of delegation because what

vides a model for the "science" of behavior, and that model

the delegated agent "does" expresses the scientist's questions.

seems to offer a way to escape the "curse of tolerance." There is

But once again this general thesis assumes its fullest expression

no one less tolerant than Freud, according to Devereux. He cre-

in the behavioral sciences. For, theoretical variables, if they are

ated a technique that proscribes nostalgia and complacency, and

to be capable not only of identifying the studied, and therefore

which expects both analyst and patient to deliberately subject

perturbed, behavior, but also that of the perturbator, obligate

themselves to the challenge of putting at risk their demarca-

the scientist to make explicit just how she is herself perturbed by

tions, how they perceive and assign to others roles and inten-

the other's behavior, which her perturbing activity has caused.

tions, being finally forced to accept that this expresses their

The perturbation of the other is in this case strictly inseparable

own unconscious fantasies. Similarly, psychoanalysis would

from the perturbation produced by the other, and identifying

then prevent the ethnologist and the sociologist from using the

the perturbation experienced by the practitioner must, there-

various methodological defenses that have allowed them to dis-

fore, become an integral part of practice. How the practitio-

tance themselves from the anxiogenic object (such as "cultural

ner "knows," or refuses to "know," how she is perturbed by the

relativism"the experiences of others have nothing to do with

other affects how she "perceives" and "interprets" the behavior

my own) or to deny that distance (expressed as the participatory

of that other. That is why the question always arises about what

observation: I disturb nothing nor am I disturbed by anything).

the observer "can bear knowing" about herself. The question of


anxiety.

And yet it is with respect to this "constructivist" psychoanalysis, which creates what it observes, that Devereux took a step

Among the behavioral sciences, Devereux emphasizes psy-

that forces me to exclaim "Curse!" Whereas the sorry, monoto-

choanalysis, of which he gives a "constructivist" version. For

nous mutilated makers of puppets that mime the autonomy of

Devereux, psychoanalysis, to the extent that it supposedly rec-

the scientific factish provoke no more than a sigh, the exclama-

ognizes that its technique creates the phenomenon it subse-

tion marks an experience of betrayal that demands elucidation,

quently explains, serves as the most accomplished example of

that confers upon the curse the power to create a problem. What

what a behavioral science should be. The analyst cannot be sat-

happened? In Devereux's case, what happened is that he tried

isfied with "This is what I perceive" as a translation of a stable

to retain for psychoanalysis, even in the constructivist version

demarcation between herself and the other: "Any analyst who

he had proposed, the privilege that Freud had attributed to it.

believes that he perceives directly his patient's unconscious,

Not only would it be the "only psychology whose exclusive and

rather than his own, is deluding himself." 10 In fact, it is always

characteristic objective is the study of what is human in man,"

"the analysis of his own disturbance (fantasy) which the ana-

it would have the means to achieve that end. And in doing so,

lyst then communicates to the patient, calling it an analysis of

Devereux turns to a definition of ethnopsychiatry that exposes

11

the patient's fantasy." And analysts' practice of the analysis of

him to the "curse of tolerance." 12

3i8

THE

C U R S E AS

TEST

THE

CURSE AS

TEST

39

It is not a question of weighing the clinical merits of what

man" and everything else, which has to do with the family, his-

Devereux called psychoanalysis or questioning his fidelity to

toiy, and culture. For Devereux, psychoanalysis can achieve its

Freud in defining the unconscious as an inseparable operator in

"universal"all-terrainobjective "because one's unconscious

the production of both analyst and patient ("only a 'preparation'

is a relatively undifferentiated function or portion of the psyche

can 'analyze' another 'preparation'").' Rather, it is a question of

and can therefore resemble that of another individual more than

learning how Devereux maintains the privilege he confers upon

can one's highly differentiated consciousness."' 6

psychoanalysis: not only the privilege of creating practitioners

This distinction between differentiated and undifferenti-

who are aware that it is the analysis of "their own perturbation"

ated, almost casually introduced into the text, is fundamental to

that they communicate to the patient, but also the privilege of

the ability to define "what is human in man." More specifically, it

allowing those practitioners to claim that this makes them capa-

is fundamental to the ability to supply a definition that, far from

ble of serving as reliable witnesses to "what is human in man."

being an integral part of psychoanalytic technique, can claim to

For, if Devereux were right, the ethnopsychiatrist, a hybrid who

provide a foundation for that technique. The fact that psycho-

has accepted the dual challenge of field ethnography and the

analytic technique obligates the analyst to believe that her own

analysis of her countertransference, would correspond to the

unconscious "resembles" that of her patient and that the analytic

psychosocial type of researcher capable of stating "I understand"

scene "prepares" the patient for confirming that resemblance

anywhere on the planet, without being herself (dis)qualified by

would not in itself present a problem. But Devereux intends to

the claim.' 4 At that point, the ecology of scientific practices could

give the presumed resemblance produced by analysis the author-

close in on itself. The modern practices of knowledge would be,

ity of an experimental fact. He wants us to recognize that what

based on their own definition and using their own resources,

the patient confirms is capable of satisfying the experimental

able to understand "nonmoderns" not differently but better than

requirement of differentiation between the two meanings of the

they do themselves.

term "artifact." It is a "technical fact" depending on a "prepa-

.And here, of course, Devereux must introduce a "requirement" that weighs on what the analyst addresses, namely, the
"human psyche." For the analyst's unconscious to be a reliable
witness of the patient's unconscious, in order for it to faithfully express the patient's message without adding or subtracting anything, the analysis of countertransference must not be a
"preparation" but a true purification. For Devereux, any deformation of the message expresses the existence of a "reserved,
frozen and inaccessible segment of the analyst's own unconscious, which cannot echo and translate the patient's fantasy."' 5
In other words, the psyche itself is defined in terms that warrant
the possibility of "dissolving" what is frozen and reveal a preexisting possibility of separation between "what is human in

ration," but this preparation also serves as a reliable witness to


the human mind as such. Occasionally, Devereux assumes a prophetic tone when stating that "We share a common humanity!"
But, if it were a prophetic statement, it should entail obligations,
and those obligations would affect both the one who made the
assertion and his or her community. The possibility that a determinate practice benefits from privileged access to "what is
human in man" entails no such obligations. Rather, it affirms
a power relationship between the one who speaks on behalf of
that practice and those affected by its requirement. It is not the
prophet but the founder of psychoanalysis as a science of behavior who requires that the "psyche" the psychoanalyst addresses
validate an approach that transcends cultural differences.

THE

3i8

C U R S E AS

TEST

THE

C U R S E AS TEST

39

When Devereux states that his "conception of normality . . .

what a healer means to the Indian. She understands the other,

is not relativistic and culture bound, but absolutistic and cultur-

but the other cannot understand her.' 8 Therefore, the miracle

ally neutral, i.e., psychoanalytic," any appearance of symmetiy

in question is unfortunately nothing but the retranslation of the

between psyche and culture suddenly disappears, a symmetry

tolerance moderns demonstrate so easily in the presence of the

that should have followed from his reference to complemen-

nonmodern whenever they claim to represent science. Here,

tarity as defined in quantum theoiy.' The existence of comple-

the possibility of such a retranslation does not strictly follow

mentary relations between the "psychic" data associated with

Devereux's premises. In fact, it occurs at the point those prem-

psychoanalytic practice and the "cultural" data produced by

ises are abandoned, specifically, whenever Devereux "forgets"

ethnology and sociology should mean that both are operational

to follow them as far as they can go. It occurs when Devereux

practices, unable as such to determine "what is human in man."

yields to the seduction of convergence between two claims:

However, the definition of the unconscious as "psychically

the prophetic-humanist claim that all humans have something

undifferentiated" suppresses the symmetry between "psychic"

in common and the professional claim that any human psyche

and "cultural." The analyst, if "well prepared," is not obligated

equally satisfies a requirement that privileges psychoanalysis.

to actualize, through the "perturbation" caused by the patient,

This may help us understand why the question of require-

the awareness that what her technique addresses as purely psy-

ments, far from being a reliable guide, has actually become

chic "fantasies" may be so defined relative to this technique

pharmacological. And when it actually becomes a poison, all

only, relative to the fact that this technique operates by exclud-

the questions raised by Devereux begin to shift, primarily the

ing from consideration cultural or social questions as such. The

meaning and scope of the notion of anxiety. For anxiety, which

analyst's "preparation" rather allows her to reproduce, for her

served as the common thread in Devereux's ecological approach,

own benefit, the power relationship between the scientist and

enabling him to understand why modern scientists so easily

her objectand to recover the asymmetiy between "modern"

stop being creative when they address behavior, then becomes

scientists, free to seek and find the appropriate culturally neu-

a poison. Anxiety is an operational concept for psychoanalysis;

tral point of view, and eveiything (here eveiyone) else defined

it is what the analysis of countertransference should elucidate.

as determined.

As a result, the behavioral scientist's anxiety does not so much

By what miracle did Devereux's ethnopsychiatrists acquire


this amazing ability? But, what price did others have to pay
for this miracle, which defined ethnopsychiatrists as mobile
and inventiveas scientificwhile they are defined by knowledge, irrespective of whether they share that knowledge or not.
We must, then, conclude that for Devereux ethnopsychiatiy
remained "white man's" science, a two-faced science: not only

concern the question of "the others"; it is formulated in such a


way that it is reduced to a question "without an object," a question that introduces no "other" aside from the researcher-analyst herself. The question raised by scientific practices has thus
been appropriated. Both its definition and its solution belong
to psychoanalysis. The operation is truly one of invagination:
the others, whom ethnopsychiatry knows how to make exist as

does the ethnopsychiatric therapist-researcher remain a psy-

humans, now find themselves inside, "at the service of science."

choanalyst even though she knows that, from the point of view

Now, heaven's rage can be expressed differently. Cursed be

of the sick native Indian she's treating, she is a healer, but she

the adult who expresses as "anxiety," to be addressed through

also claims that her analytic practice enables her to understand

analysis, the effect of the child's outrage upon her.

ANXIETY AND

FRIGHT

325

Although anxiety is an integral part of Freudian technique,


outside that technique it does not have, Nathan notes, any claim

21

to a universality that would give the analysis of countertransference access to "what is human in man." However, Nathan

Anxiety and Fright

did encounter another concept that is important in traditional


therapeutic technique and thought, the concept of "fright." Yet
and this is what I find interestingNathan does not claim that
"fright" is the "true" universal, which could be used to establish
a science of "what is human in man." For him, it is part of an
"interactive etiological system that brings about multiple psychotherapeutic procedures, which are also always interactive." 1
In short, it belongs to a technical definition of both illness and
healing.

When Tobie Nathan, Devereux's intellectual heir, reinvented

For Nathan, Freudian psychoanalysis emphasized anxiety

the field of ethnopsychiatiy, he also challenged the postulate of

over fright because the latter "brings with it an infinite array

the "universal human" implied by analytic theoiy and the refer-

of external causes, whereas anxiety, more austere, purer, justi-

ence to anxiety as an all-purpose anthropological tool. In doing

fied a sophisticated technique in which thought is eveiything,

so, he shifted the issues of ethnopsychiatiy. Devereux's anxiety

similar to the neuroses whose treatment psychoanalysis would

allowed him to follow the way the landscape of scientific prac-

devote itself to."2 Consequently, the theoretical crux of the dis-

tice was highlighted by the contrasting question of the "per-

tinction between anxiety and fright would be "primarily clinical,

turbations" that express what scientists require from what they

even technical. To acknowledge fright as a central psychic affect

investigate, the tests to which they subject what they question.

would have led to the conclusion that some other' has intruded,

Nathan has no ambition to unify the landscape of modern sci-

has influenced or modified us, possibly even caused our meta-

entific knowledge. He is interested in constructing an approach

morphosis. This also would have meant admitting that the psy-

that prevents researchers from disqualifying nonmodern psy-

choanalyst, through his presence aloneor rather through the

chotherapeutic practices under the pretext that they cannot be

psychoanalytic settinghad triggered immense fright leading

defined as "scientific techniques." It is not analytic technique

to repeated transference events. . . . For, the essential fright is

or anxiety as such that he is challenging, but their claim to ben-

that the truth of what I perceive, of what I feel, of what I think,

efit from a special relationship to the human "psyche," beyond

resides in an other. But if truth is found only in that other, if

cultural differences. That is why Nathan's challenge will lead us

I have no way of knowing how the other is frightened as well,

to ask about the ecology of our knowledge, the way in which we


require of all techniquesours and others'that they be evaluated in terms of a "universal" reference that selects those that
are rational and explains their efficacy.

324

then the entire effort is merely a new form of fright, and not
therapy." 3
When seen from the point of view of the fright caused by the
analyst, psychoanalysis is not disqualified. For we are confronted

36

ANXIETY AND

ANXIETY AND

FRIGHT

by the question of the obligations of the "frightening" therapist,


including those of the Freudian therapist, even if in Freudian
therapy the notion of fright is not explicitly introduced. More
specifically, the Freudian therapist would be obligated to dissociate her practice from its scientific claims and assume responsibility for managing a technique a unique active dimension of
which is that it does not make explicit the fright it works with.
This is a practical responsibility from which the analysis of the
countertransference will not release the therapist.
Starting from the question of the therapist's obligations,

FRIGHT

327

existence. For Nathan, relating illness to other universes is itself


a component of a technical approach, for it causes interest in the
patient to shift and to mutate: "Whenever you are a patient and
someone does something foryou, the entire group benefits from
this and is conscious of this fact." 5 The meaning of the message
associated with the illness now concerns the entire group. The
patient has become a "point of intersection" between the everyday world and the "other universe" from which the message
originated.
The obligations involved in the process of divination have

Nathan was also able to contrast processes of divinationread-

the abstract character I associate with constraints, and it is the

ing coffee grinds, for exampleassociated with therapeutic

production of abstraction in the strong sense of the term, exclu-

practices with diagnostic procedures, such as the Rorschach

sive of truth or fiction, that seems to be implied by the process

test, which is part of a purportedly scientific practice: "In fact,

(see Cosmopolitics, Book I, "The Science Wars"). Lucien Hount-

[readingl coffee grinds provides a considerable methodologi-

pakin, discussing the Yoruba babalawo, or "father of the secret,"

cal advantage over the Rorschach test. Naturally, it is also a kind

writes that the "active word" he gives his patient "acts as a spe-

of projective test, but one undertaken by the clinicianwho is

cific asset, a word unlike any other. The babalawo is a true artist.

obviously, in this case, a seer rather than the subject. And since

He has to gather elements scattered throughout the culture, he

it is administered in the presence of the subject or an object rep-

has to cover them with 'common words,' which are immediately

resenting the subject (a photograph or an article of clothing), the

accessible, and slip it all into a formula that resists time and

reading of the coffee grinds can only provide information about

space so that, even ten years later, the patient can still use that

the state of the relationship between clinician and subject, not

word to enrich his existence." 6 To resist time and space and pro -

the hidden hypothetical nature of the so-called subject. So, the

duce a renewed relevance in circumstances that have yet to be

reading of coffee grinds would, strictly speaking, be a technical

determined, few philosophers have dared offer such a beautiful

procedure intended to force the clinician to speak only about

definition of the construction we call abstraction.

the interaction that he has established with a person, and consequently to produce usable clinical material."

Divination must be described in terms of obligation: the


obligation of the clinician-seer to "produce" "material" under
conditions in which she cannot yet "possess"in the customary
sense, as well as the much more disquieting sense the term may
connote, in this quite similar to the pharmakonknowledge
about the patient. It is also the obligation to expose oneself, to
become the site of a dramaturgy that brings other universes into

I want to point out a crucial element in Nathan's writing on


"nonmodern" therapeutic practices: it is as if it were imperative
for these therapeutic techniques to confer upon the illness a significant value based on a strategy that creates obligations where
modern practices have requirements. 7 The multiple "supernatural" universes that such practices imply, confirm, and bring into
existence do not satisfy the requirements associated with modern proof, but they are therapeutically relevant. They refer to the
therapeutic process as one that unravels hidden intentions and

36

ANXIETY AND

ANXIETY AND

FRIGHT

causalities through which a patient can be "produced," capable


of interpreting what is happening to her and, if need be, of joining a "real group" to which her trouble means she belongs and
which creates specific obligations for her.
There is nothing romantic about such "supernatures" [surnatures]. They are not a spiritual supplement that can be used to
reenchant a silent world, a source of heightened inspiration, a
fabled return to a lost innocence, a reconciliation with a nature
in which birds warble the words of gods. The gods, djinns, and
ancestors of the therapists described by Nathan do not speak
in the transparent immediacy of a lost paradise; their message requires a technician, technical and constructive thought.
They belong to the technicians practice because this practice is
directed not toward the satisfaction of requirements but, like his
patients, their relatives, theirfamily, and at the same time as them,
toward the decoding of obligations. Seen from the modern viewpoint, these multiple "supernatures" can only be understood in
terms of beliefs that must be tolerated. Their therapeutic efficacy may even be recognized by modern therapists, the way
Plato's heirs regretfully recognized the art of sophists capable
of influencing a crowd by their persuasive speech. But it is precisely at this point that the possibility of an escape from our own
judgments arises. It is not a question of knowing whether Tobie
Nathan's interpretation is the "correct" one, if it is true, if it "corresponds" to what healers do in practice. 8 The fact is that this
interpretation, which is inseparable from an ethnopsychiatric
practice initially created to meet the psychotherapeutic problem
presented by populations of immigrants in France, forces us to
consider what a real "technique of influence" means, when distinguished from the "effects" of influence, or suggestion, that
we are all supposedly able to produce at all times, under all circumstances, whether we want to or not.
The work of Tobie Nathan and his colleagues allows us to
appreciate the challenge of "fetishes" and practices we disqualify

F R I G H T 168

329

as "fetishistic." Nathan does not offer us access to such practices


"in themselves," in their authentic truth. Rather, what he presents are such practices characterized in a way that forces us to
think. And primarily he forces us to think about the contrast
between the ideal fabrication of the analyst, in the sense that
Devereux has effectively taken from Freudthe "purification"
through the analysis of countertransference of what blocks the
analyst's ability to listenand the "fabrication" of the nonmodern therapist.
That the psychoanalyst and not the medical psychiatrist plays
a role in this contrast indicates a common trait: in both cases the
therapists are supposed to "have been ill" and to have followed
a path that makes them capable of practice. But this common
trait is limited, especially given that nonmodern paths of initiation are not only multiple but also largely secret. Therefore,
what forces us to think is not any particular element of the process, whether it consists of objects, words, songs, gestures, or
rhythms. It is not a question of conquering knowledge that the
process hides from us, or discovering the way in which it orients or guides practitioners. Being forced to think has nothing
to do with feeling that we have the right to make explicit what the
other might know "implicitly." On the other hand, what might
force us to think is the contrast between the analytic purification whose ideal is so much a part of our traditions of truth, both
ascetic and scientific, and the contrived multiplicity proposed
by traditions we refer to as "fetishistic." The possibility of suggestion, of influence, is for us a key, a quasi-anthropological
characteristic, and psychoanalysts are obsessed with the question of knowing how not to influence, how not to "deform" the
message addressed by the patient's unconscious to that of the
practitioner. In contrast, nonmodern technique turns "influence" into a sophisticated question. The psychoanalyst follows
an apparently modern path, contrasting "influence" and "reason," and in fact the "reasons" of the unconscious do not have

36 A N X I E T Y

33

AND

FRIGHT

any other definition: they are what take precedence to the extent
that the analytic scene has been purified of all the influences

ANXIETY AND

FRIGHT

169

produced along "with the patient" but elsewhere, among competent colleagues. 9 More generally, whenever a patient meets a

that might have been presented as reasons. It appears that non-

modern therapist, she implicitly deals with this group of compe-

modern practitioners, however, must furnish both their thought

tent colleagues who maintain the rational adequacy of technique.

and their practical environment with any number of appliances


if they are to become capable of "fabricating" thought.
How can we characterize an "influence" that is "produced,"
actively composed of heterogeneous elementssupernatural
intentions, objects, words, gestures, family groups, various
materials, natural and cosmic, ancestors, the attacks of sorcerers, obligations betrayed or ignored? Can we use words such as
belief, suggestion, effectiveness of the symbolic, which we employ
to refer to "others" and to that part of ourselves that, as Pascal
wrote, "reason does not know"? Or, on the contrary, are we obligated to recognize that such expressions are so many "master
words," which can be used to pass judgment without having to
encounter or experience, which can be used to avoid turning the
practices of others into witnesses of a problem that is liable to
"frighten" us, that is liable to call into question our own modern
requirements?
The contrast is likely to call into question what our master
words define as the norm. If we accept that what nonmodern
therapeutic practices do not "lack" is proof, the imperative that
modern psychotherapeutic explanation must refer to, that would
prove their adequacy assumes a different meaning, creates
another, unexpected, dimension of the contrast. It is true that
the Freudian ambition to merge therapy and research, that is,
to enable therapy to "prove" its own adequacy, has always fallen
short whenever it confronts those familiar with experimental
proof. But let us assume the correctness of Freud's ambition: the
successful transformation of unconscious conflicts, resistance,
and repression into "factishes" endowed with an autonomous
existence that forces every researcher involved with the psyche
to take them into account. Such success would not have been

For Nathan, the imperative of proof requires that all "modern"


therapies fabricate illnesses defined by symptoms that must be
"welded to the individual." The individual then finds herself
isolated and categorized, which is to say, subject to the knowledge of the therapist and his colleagues. "The psychopathological categories that psychiatristsas well as psychoanalysts and
psychotherapistsuse to classify their patients are never based
on real groups. Have you ever heard of groups of people suffering from obsessive compulsion . . . gathered together in the
same place, acknowledging that they all sharewho knows?a
common ancestor, before subjecting themselves to the same
therapeutic ritual? . . . Of course not. Because psychopathological categories are concepts that separate, only 'regrouping' individuals statistically." 10
What produces this contrast, which calls into question the
requirements associated with proof? Somewhat unexpectedly,
I feel it leads us to that showcase of our rationality, the laboratory. Until now, the laboratoiy has been described primarily
as a place where the requirements that our "modern" factishes
must satisfy, where the invention and discovery of how a being,
an electron, a molecule of DNA, or a Pasteurian microorganism
satisfies the requirements that define its mode of experimental
existence are celebrated. The laboratoiy, however, is also a place
saturated with obligations. How does a physicist today know that
electrons exist? If we question her critically, we may come to the
conclusion that she "believes" in the electron, and the critical
inquirer will then be able to gloat over such "fetishistic" belief.
But if we abandon the crude dichotomy between knowing and
believing, we are forced to conclude that physicists today are
nothing without the electron (or molecular biologists without

ANXIETY AND

36

FRIGHT

DNA), for the majority of devices that populate their laboratory


assume its existence. If the field is one in which we can state that
the present is child of the past, in which the practitioner would
be disarmed, powerless, if not authorized by the past, the field
in question is that of experimental practice. If there is a practice
where thought must not be expressed solely in terms of requirements but also obligation, where the beings to whom practitioners refer have as their primary function to force them to think,
to participate in the production of thought, it is experimental
practice. And if there is a place where the beings created have
the means to obligate those who create them because they must
appeal to such beings to confer meaning on the technical apparatus without which there would be no thought, it is indeed the
laboratory.
The laboratory is the place of proof, and its possibility
depends on the fact that the questioned being satisfies requirements presupposed by experimental proof. If the electron is an
experimental being, if its existence has been proven, it is not only
because it has managed to satisfy the requirements and proofs
that were specifically addressed to it, but primarily because it
satisfied the general condition on which the veiy possibility of
proof depends: the electron is indifferent to the proofs we subject it to. A rat "knows" when it is in pain, while the many roles
assumed by the electron in the laboratory and in innumerable
technical appliances seem to confirm that it "knows" nothing
about itself in Devereux's sense of the term. This general condition means that if the questioned being, whether a human or a
rat, cannot be considered indifferent to what it is forced to
undergo, the distinction between the two meanings of an artifact,
which must be made in experimental laboratories, is pointless.
The "fact" about the rat in Skinner's cage will always be dependent
onthecageinwhichitwasproduced.Itcannotbeusedtoconstruct
an autonomous factish that both exists "outside the laboratory"
and explains what is happening to it in the laboratory.

ANXIETY AND

FRIGHT

170

From the point of view of the requirements of proof, the idea


of "behavior" is disruptive. But the question of obligations can
be used to transform the opposition between a "real laboratory"
and a "pseudo-laboratory" (such as the Skinnerian laboratory)
into a contrast. What characterizes such pseudo-laboratories
then becomes the fact that not only the scientists who work
there but also their "objects" "serve science."
Even those who force rats to run around mazes know just
how much the relationship with the rat "counts," as if the rat had
to "feel obligated" to carry out the requested performance. To
what extent did the "torturers" in Milgram's experiment agree to
administer increasingly painful electric shocks to their victims
simply because they felt obligated by being of service to science? And didn't the reference to being in the service of science
and the legitimacy of scientific experimentation also allow the
experimenter to require that his subjects reveal to science, but
also to themselves, their "potential" as torturers. The "behavioral laboratoiy" is not only the place where both meanings
of the term "artifact" can no longer be distinguished; it is also
where the need to distinguish between them is expressed by the
confused and terrifying submission to obligations that produce
the behavior of the experimenter as well as that of her subject."
In other words, it cannot be the place where our "modern factishes" are produced because all agents there are defined in
terms of their service to science.
The practice described-constructed by Nathan, and the
"supernatures" it incorporates, are not unique because those
supernatures are vectors of obligation. So is science in the
pseudo-laboratoiy. It is unique because those obligations force
both therapist and patient to think, as well as the entire group
affected by therapy, and because they are recognized and cultivated as such. In other words, although we can speak here of
a "laboratoiy of influence," the important difference is that,
unlike our "behavioral laboratories," such a laboratory does not

36

ANXIETY

AND

FRIGHT

create an obligation that incorporates a clandestine and uncontrollable ingredient, does not maintain the pretense, in the
name of science, of a well-defined differentiation between the
one who knows the questions to be asked and the one who has
to answer them. If we must continue to speak of belief, our own
requirements turn against us. They identify us as "believers,"
as those who believe they are released from the obligations that
would qualify others.
How can we reconnect with the obligations of our own
practices, when the requirements they claim to extend appear
to turn us irresistibly into believers? That this question can be
asked expresses the fact that the possibility of escaping the curse
of tolerance is no longer a matter of having a bad conscience,
but a practical question. Tobie Nathan, within the framework
of the ethnopsychiatry that he renewed, suggests a practice, and
perhaps a key. The co-therapists brought together by the process undergo a strange trial: they must participate in, or themselves engage in, the construction of "active" words and objects
required by the "nonmodern" tradition to which the patient
(originally, an immigrant) in their care belongs, whether the
patient wishes to ignore it and adopt European categories or not.
Here, analysis of the countertransference ceases to be a panacea,
a "universal" remedy: when undergoing this trial, it is no longer
possible to associate the anxiety experienced by the co-therapists with general, undifferentiated reason, such as the fantasy
of a "magical," "all-powerful" relationship. Rather, the process
obligates them to confront what we call "beliefs." A co-therapist
is cursed who seeks to maintain a judgmental relationship with
respect to those "magical" gestures and words, accepting them
only as conditions of a relationship she must establish for the
"good" of the other. For, in so doing, she fails to acknowledge
the frightening power of intrusion associated with the supernature to which those gestures and words refer.
The "fright," which here replaces anxiety, does not imply

ANXIETY AND

F R I G H T 171

335

conversion. 12 The ethnotherapist is not frightened herself, does


not attribute to the active object the power to act on her the way it
acts on those belonging to the group that produces it. It does not
express an obligation to "belong" to a given group but, rather,
the obligation to recognize, in a way that is not a form of reflection, one's own belonging. The "perturbation" Devereux spoke
of no longer refers to a psychic "core" common to all humans
but a cultural core, one that produced modern humans in such
a way that they recognize and deplore "fetishes" wherever they
find their manifestations. As for the curse, because it is related
to a practice, it loses all its prophetic character and is relative
only to the creation of values that engage the practice of ethnotherapy. Anyone who feels free to "mimic" "magical gestures,"
all the while remaining aloof, is as misguided on the ethnotherapeutic scene as the experimenter who feels free to "cheat" in
the laboratory or the field scientist who feels free to attribute to
the causes she identifies the power to transform narrative into
deduction. For reasons that concern her but are unimportant
for practice, she is simply unable to participate in the production of thought this practice assumes.
So, we end up with certain ingredients of the definition of a
new psychosocial "type" of researcher, a researcher whose practical field of belonging actively prevents maintaining an opposition between modern knowledge and "traditional belief." For
Nathan, this researcher alone would be capable of sharing in the
realization of Devereux's grand projectthe creation of a finally
scientific psychopathology. This psychopathology wouldn't tiy
to require a definition of "illness" that justifies the privileges of
a specific healing practice, one that satisfies the same categories
as the definition. If eveiy psychotherapeutic practice "creates"
t he illness it heals, that is, if it incorporates it into an operational
procedure, where words, actions, meanings "produce" it so they
can act on it, a scientific psychopathology must engage with practices, not patients. It would then require modern researchers to

ANXIETY AND

36

FRIGHT

recognize as their "therapist colleagues" those who have mastered those practices.
However, the path followed does not stop at the question of
a psychopathology that is finally scientific. Rather, this question
leads to others, which are far more speculative. If the Yoruba
"father of the secret" must be recognized as a colleague, such
recognition cannot be conceived unilaterally or we would be
back at the situation we hoped to escape from, the one in which
we always "know better." For, the most extreme asymmetry
would arise wherever there was a possibility of symmetry. However, this "colleague" can't be separated from the group to which
he belongs. The question then becomes one of learning how
"Yoruba colleagues" can be addressed in a way that confirms
their belonging to the group that authorizes them to speak, by
making those groups "present." A scientific psychopathology,
therefore, raises the question of a "group-to-group" relationship, and the threat of asymmetry now arises from the radically
different makeup of the groups in question. In other words, the
problem now becomes "how to live with the Yoruba." Once again
the question of "modern practices," including ethnopsychiatry,
cannot be limited to its own concerns. We are not yet done with
obligations.
Until now, I have used the term "culture" without defining
it, although the term is subject to multiple, antagonistic definitions. I want to address that definition because of the urgency of
the question of "how to live with the Yoruba." I'd like to try out
the definition proposed by Tobie Nathan, for what he defines is
not an entity but a problem: "Culture tries to resolve two problems: closure and transmission. How can a group be enclosed
so that it is impermeable to others, and how can this closure
be transmitted to the next generation? To resolve this technical problem, every culture makes its own choices."' 3 The closure
Nathan speaks about is not physical but logical, in the sense that
logic is, obviously, inseparable from practice and cannot be

ANXIETY AND

FRIGHT

172

made explicit and formalized in logical terms.' 4 Moreover, this


closure, which characterizes what Nathan refers to as "ethnic
groups," is explicitly associated with the question of exchange.
"In Cairo there were thirty, maybe forty, ethnic groups who paid
close attention to their reciprocal definitions: a Copt from one
place was not the same as a Copt from another place. An Orthodox Greek was not a Catholic Greek, and so on. This resulted in
clearly defined ethnic groups; but these groups were very familiar with one another and felt that it was a matter of their personal, individual survival to preserve those cultures because that
was the condition of exchange."' 5 Here, culture has nothing in
common with a "closed" identity, in the static sense that is often
associated with the notion of autopoiesis.' 6 Culture is part of a
technical problematic that is strictly ecological in nature: closure and the transmission of closure as "condition of exchange."
What's more, it cannot, as such, be subject to a theory. Each
mode of closure and transmission constitutes a unique solution
that creates its own constraints, problems, requirements, and
obligations in terms of exchange. In short, it creates the practical identity of what is called "exchange."
It is important to bear in mind that culture as it appears
in the strictly ethnopsychiatric problematic, where the question of healing arises, and culture the way an ecology of practices presents the problem are distinct. The culture to which
the therapeutic process actively refers is relative to this process
and to the "disorder" whose meaning is to be constructed and
situated. The process and those who speak on its behalf are not
capable, as such, of representing what culture in general is and
what it is capable of. Their evidence relates to the way in which
it is tested and implemented through the practice of therapy, the
way in which it participates, as a resource and ingredient, in that
practice.' 7 Consequently, it is in technical rather than scientific
terms that the question of culture arises in ethnotherapeutic
practice. The definition of culture is not, and should not be, the

36

ANXIETY AND

FRIGHT

same whether it relates to the "exchanges" between cultures or


to therapeutic practices.

ANXIETY AND

FRIGHT

173

authority. This ability may be an individual trait, but requirements cannot depend on individual character. From now on, 1

But the perspective of the "finally scientific psychopathol-

will refer to those able to resist as "practitioners" in the broad

ogy" Nathan suggests not only presents the problem of obliga-

sense, meaning people engaged in some activity or role that

tions, to which the question of "exchange" respondshow to live

entails obligations and requirements. Didn't more reliable

with the Yorubait also provides a partial answer to the ques-

knowledge about what a housewife is find support for its condi-

tion put forth here: how can what I globally refer to as the social

tions of possibility in the work and commitment of feminists,

sciences be practiced without becoming subject to the curse of

for whom such questions mattered? Likewise, a possible start-

tolerance? Such a perspective provides the means to approach

ing point for knowledge about those we call drug addicts can be

the practices of these sciences, like all the others, in terms not

found in the confrontation with associations of "unrepentant

only of obligations but of requirements. We can now ask about the

drug users" who challenge, practically, politically, and concep-

requirements associated with a scientific practice because those

tually, the representations we have constructed about them.

requirements will no longer be focused on the question of proof.

At the beginning of this book, I stated that the question asked

Their construction follows from the obligation I proposed ear-

involved two entangled populations: "scientists" and "media-

lier: resisting the challenge of the "curse of tolerance."

tors," the latter referring to the authority of forms of knowledge

Why would a science be possible "with the Yoruba colleagues"

that claim to be "scientific." If my hypothesis is correct, if sci-

(and others) but not with patients? Because, unlike patients,

entific practices in the "social sciences" require that what they

who are there to explore, through a technique of influence, the

address has the practical ability to put the researcher at risk, its

obligations entailed by their illnesses, such "colleagues" will not

first consequence is the problematizing of this entanglement.

be (overly) influenced but will be fully capable of putting at risk

We must conclude that when practitioners address what is com-

the way in which their practice is described. In what sense is a

monly called a "socially disadvantaged" population, they can

"social" science possible? Is it conceivable to create a psychol-

no longer refer to the authority of science because the categoiy

ogy of the "housewife," the "adolescent," or the "TV viewer," for

"socially disadvantaged" excludes the construction of knowledge

example? How, in such cases, can we claim that what psychology

that would be scientific. But when an intervention has something

addresses is capable of putting at risk the arguments that refer

to do with science, when it: is not to be cultivated as a technique

to it? TV viewers, adolescents, and housewives are categories

of influence, it will always mediate between possibly conflict-

already defined as the targets of techniques of influence. Don't

ing practices, each of which is capable of presenting itself in a

scientists take advantage of this social construction to secure a

positive manner. Mediation will then imply, like the question

stable difference between those who ask questions and those

"How can we live with the Yoruba?" the creation of the "practi-

who bear witness, through their beliefs, opinions, and pre-

cal identity of what will be called exchange." Which means that it

sumptions, to the influence they are exposed to?

is urgent that the question of the difference between scientific

Just as laboratory practitioners require what they question


to be capable of the purification needed to produce reliable evidence, practitioners of the social sciences, in order to construct
reliable knowledge, need to encounter people able to resist their

practices and technical practices be made to exist.

THE

POLITICS

OF T E C H N I C A L

INVENTIONS

341

an "event," none was able to situate us as its heirs, whether we


extend its consequences or resist it. In other words, the category

22

of "technique" remains de facto available for mainstream use,

The Politics of
Technical Inventions

"science" from "technique"? To establish the scope of our dis-

regardless of the question of modern and nonmodern.


How can we approach the question of what distinguishes
cussion, we must first recognize the category, itself well defined,
of "scientific technique." Perhaps the simplest example of
scientific technique is the cycloidal pendulum invented by
Huygens. From a scientific point of view, the law of isochrony
associated with the pendulum when it is used as an instrument
to measure time, the fact that the period of a pendulum (at uniform gravitational acceleration) depends only on the length of

It may appear surprising that I feel the distinction between


scientific and technical practices needs to be explicitly constructed. In fact, and even though the word science refersas it
does hereto the "modern" sciences alone, the distinction is so
ambiguous that some feel it should be abandoned altogether for
a more comprehensive concept: "technoscience." However, to
treat technique as distinct is crucial if we are to resist the effects
of fascination associated with discourse about the inexorable
technoscientific redefinition of the world. To begin this construction, I'll begin with a minor asymmetry between science
and technique. In The Invention of Modern Science and throughout
Cosmopolitics, I took the risk of accepting the difference between
the so-called modern sciences and other knowledge practices.
But I did so on condition of presenting this difference, of making it present, in a way that differs significantly from the way it
is ordinarily introduced. It is much more difficult, however, to
talk about the "invention of modern techniques." Attempts have
been made to define what a modern technique might be, that is,
to politically define the "technician" in the community in the
same way the "scientist" was defined at the moment the modern sciences were invented. However, no such attempt marked

34

the cord, is valid only for oscillatory movements whose amplitude is small. So isochronism is an approximation. Huygens's
technical invention is fhe invention of a device that transforms
approximate isochrony into exact isochrony. This device consist
of nothing more than two blades attached on either side of the
pendulum's anchor point in such a way that, depending on the
amplitude of the pendulum's motion, a portion of the cord will
contact first one, then the other of the two blades. As a result,
the motion of the pendulum is modified (without friction) and
the apparatus behaves "as i f " the length of the pendulum's cord
varies as a function of its distance from the vertical. Of course,
the shape of the blades (cycloidal) was calculated using the scientific law of oscillation in such a way that the modified motion
is exactly isochronous, regardless of the amplitude of the pendulum's oscillation. This shape, therefore, embodies a mathematically well-defined transformation of motion. We have here
a perfect example of technique as applied science. The general
law of pendular motion is known and guides the required modification of pendular motion in such a way that the pendulum can
become a reliable instrument for measuring time. Once this is
done, the technically reinvented pendulum makes exact what

342

THE

POLITICS

OF T E C H N I C A L

INVENTIONS

the scientific pendulum expressed only approximately.


Note, however, that the above example is a bit too perfect.
The ideal transparency of the invention of the cycloidal pendulum is matched by its limited scope. This pendulum, if placed
aboard a ship, will be useless. 1 It requires a protected environment, as is true of most laboratory instruments whenever they
continue to operate exclusively in their place of birth, in the

THE

POLITICS

OF T E C H N I C A L

INVENTIONS

343

primary characteristic is their heterogeneous and contingent


nature. That is why the cycloidal pendulum is a superb example
of applied science but a poor example of technical invention. It
satisfies a single obligation associated with the practical values
of its users; it makes isochrony "really" true. But it disregards
all the other requirements a "timekeeper" worthy of the name
should satisfy.

hands of "competent" users, capable of using them and evalu-

The "modern technical laboratoiy" (which includes all ser-

ating the conditions, limits, and meaning of their reliability.

vices, including market research and legal services, that are

Strictly speaking, a technique is "scientific," that is, directly

present during the design of a device) escapes the obligation

authorized by scientific practices, only within this framework.

that identifies the scientific laboratory, the distinction between

When the instrument migrates to "foreign" labs (for example,

fact and artifact. It produces something new rather than reli-

when "nuclear magnetic resonance" migrates to hospitals), the

able witnesses. However, it must satisfy a veiy large number of

qualifier "scientific" incorporates a value judgment that hides

disparate obligations expressing constraints that, in one way or

the reinvention that gave it the power to leave its place of birth.
Although, strictly speaking, a scientific technique above all
celebrates the satisfaction of a requirementwhat is called a pendulum ideally satisfies (whenever friction is made to approach
zero) a mathematical definitionthe reinvention that gives a
scientific technique the power to migrate must be expressed in
terms of obligations. Understood in the practical sense of creating links, rather than the speculative, rhetorical, or political
sense, the ecology of scientific practices is invented and negotiated primarily by means of such migrations. What links physics and biology is, first and foremost, technique: spectrography,
radioactive marking, nuclear magnetic resonance, and so on.
In this case, the obligations that technique must fulfill in order
to migrate relate to the practitioners who accept it in their lab

another, will have to be integrated into the definition of what


is produced. Under what conditions will the instrument function, what qualification will be required to use it, what sort of
reliability will be necessaiy, what price can be charged, are
there existing patents that might cause a problem, what safety
requirements or standards will have to be respected? All these
questions and many others share, along with the science that is
supposed to "explain" it, in the process that brings a product or
technical device into existence. The modern opposition between
rationality and opinion has no authority here, at least not automatically. Eveiything depends on the evaluation of power relationships. A device can incorporate something judged to be a
mere habit, or risk taking the opportunity to alter that habit.
To speak of obligations here is to underline that none of the

as a reliable ingredient for argumentation or proof. But when-

constraints that share in the invention of a technical device has

ever migration is toward places where answers are developed to

in itself the power to determine how it must be understood.

questions that are not scientific but have been raised by other

For example, it is the "technical laboratoiy" that has to deter-

groups, or on their behalf, the difference is indicated by a radi-

mine the meaning it will confer on the knowledge developed in

cal reinvention. In this case, the obligations that preside over

"scientific laboratories." A scientifically established "property"

such reinvention do not only involve a change of users. Their

can be exploited as well as avoided. The technical design then

342

T H E

POLITICS

OF T E C H N I C A L

THE

INVENTIONS

ensures that it has no consequence on how a device functions.


In other words, the "technical laboratoiy" is obligated but not
determined. It transforms obligations into determinations
for each particular device, each determination then assuming a meaning and being articulated with others. The implications of the social, the legal, the economic, the scientific do not
preexist the production of the technical device as such. If that
device, once produced, can be described in terms of those different "points of view," it is because this production was also the
invention of the meaning to be attributed to what matters from
each of those points of view and the coherent articulation among
those different meanings. 2
Why, then, speak of a "modern technical laboratoiy" when
the operations of disqualification, of opposition between the
"rational" and the "illusory," the search for a fixed point from
which the distinction between "objectivity" and "rhetoric" could
be stabilized, have no authority there, when some of the obligations that participate in the production of a technical assemblage
openly display that what they refer to has no fixed meaning,
when the meaning of every constraint can have the instability
of the pharmakon: poison or remedy, obstacle or opportunity?
Whereas it is important to restrict the categoiy of "scientific
technique," which provides technique with a legitimacy that it
would obtain from science, the categoiy of "modern technique"
must be maintained because the laboratory I have just described
is emphatically not a general model. It is "modern" in two
respects, one of which is constitutive, the other opportunistic.
The technical laboratoiy as we know it is modern primarily
in that those who are active there must, ideally, satisfy an obligation that identifies them as modern. Each represents a dimension of reality, which is to say, the obligations that ensue, but
none can claim to transform an obligation into a condition. They
must all be ready for operations of reinvention, for the muta3

tions of meaning by which the device will be invented. In other

POLITICS

OF T E C H N I C A L

INVENTIONS

344

words, the "representatives," or spokespersons, must strip what


they represent of any claim to constitute, in itself, a "cause"
or "condition" that must be respected as such. They have to
"abstract" themselves and what they represent from the ensemble of relationships of reciprocal capture that matter for them
as practitioners in order to make it available for entering into
new, concrete arrangements. The modern technical laboratoiy,
therefore, requires a "doubly modern" world: a world capable
of providing it with "technicians" who actively and deliberately
proceed with the reinvention of what they represent, and a world
capable of accepting the reinvention of meaning and values that
a new technical device or product entails and will try to instigate
as soon as it leaves the laboratoiy.
Where the environment consists of other technicians, lawyers, standards auditors, even consumer associations with access
to "counter-laboratories," the reinvention-betrayal is limited.
Just as the properties that make gold gold have been multiplied
and become more exacting throughout a histoiy in which alchemists and testatores were at odds, the history of standards, legal
requirements, and sales criteria constitutes the translation of
multiple histories in which were negotiated the spectrum of differences between constraints, which technical invention can
attempt to reinvent, and conditions, which it must apply as is.
The difference between constraint and condition, therefore,
expresses the power relationship between the laboratoiy and an
environment that has the power to eventually impose a standard
whose meaning cannot be reinvented, that is to say, the power to
ensure that the laboratoiy satisfy clearly determined demands.
However, whenever the environment can be assimilated to
the "public" of consumers and users, the technical laboratoiy,
as we are familiar with it, can be said to be modern in quite a
different way. It exploits the opportunity offered by the division
between rationality and opinion. Even the demands of an industrial enterprise may be ignored as incompetent "dreams" when

342

THE

POLITICS

OF T E C H N I C A L

THE

INVENTIONS

(as was the case in the early days of computer technology) it has
no countertechnicians able to examine a technical proposal.
The technical laboratory then makes active use of the rhetorical
resources of modernity to claim that what it produces is accepted
as such, to silence or disqualify those who do not have the power
to force it to consider as obligations the conditions, requirements, and distinctions they would like to see recognized. It is
then that the opaque argument of the "requirements of technical
rationality" makes its appearance, or that a device will be presented based on the science that is supposed to explain it. At this
point, social or economic "needs" with a given identity, which
the device satisfies and which justify it, will be brandished.
Stories flourish about rational, scientific knowledge becoming
able, because of the discovery of the laws to which phenomena
are subjected, to use those phenomena to satisfy those needs.
Here, the "modern Constitution," in the sense Bruno Latour
calls it into question, becomes functional, lending its syntax to
the statements that shackle our thinking about hybrids by referring them to two agencies supposedly capable of being defined
separately: nature and society.4.
This second aspect of the modernity of technical laboratories
obviously expresses a major political problem in which what we
understand as "democracy" might be gambled or lost through
derision. Unlike practices organized around the requirement of
proof, technical practices do not, as such, contravene the obligations of democracy, and the possibility even exists of a convergence between these obligations and the requirements imposed
on invention. The modern technical laboratoiy, if it is actively
stripped of the rhetorical resources of modernity, communicates directly with a cultural-social-political question. How can
those who are affected by what is being produced be "invited" to
participate in its production? How can they become concerned
parties, multiplying questions, objections, and requirements?
Doing so separates what I have distinguished as "constitutively"

POLITICS

OF T E C H N I C A L

INVENTIONS

346

and "opportunistically" modern. It complicates the technical


invention, certainly, but in a way that does not violate its mode
of functioning. To this extent, what I call the "modern technical laboratoiy" is, as such, an exemplary site of experimentation
for what Bruno Latour has called, generically, the "Parliament of
Things." 5
Needless to say, the perspective of the Parliament of Things
in this modern reading is Utopian. It is nothing more than an
empty dream if it does not function as a diagnostic vector for
what makes it merely a Utopia, and as a learning opportunity
for resisting what today opportunistically frames our world.
The multiplicity of practices that gather around issues and must
invent their interrelations, implies an obligation to actively
resist the hierarchy of knowledge, to make an active distinction between scientific and technical inventions, to engage in
a deliberate process of clarifying the requirements and obligations of each practice. It especially implies the creation of new
types of practices, of "interest" groups capable of promoting
ignored dimensions of issues and situations. A single condition is placed upon participation in the Parliament of Things.
Eveiyone may introduce themselves in terms of the requirements their practice brings into existence. But in a hypothetical
trustworthy and reliably functioning Parliament, everyone must
also agree that those requirements be expressed as obligations
during the process of deliberation/invention. No one can introduce themselves by establishing conditionstake it or leave it
from which the possibility or impossibility of agreement would
follow.
"Planck" would be welcome in the Parliament of Things,
with the conservation of energy, which would apply "even to
Martians." But the unified view of the world he defended would
not resist the challenge of having to meet those it disqualifies.
And the historical analyses of "Ernst Mach," who embedded the
"energy" factish in the practices and questions of technicians,

342

THE

POLITICS

OF T E C H N I C A L

INVENTIONS

THE

POLITICS

OF T E C H N I C A L

INVENTIONS

348

the event in which those practices are gathered and do not dis-

In short, they must cooperate with the invention even when they
complicate the process.

qualify factishes as simple logical generalizations. Galileo's

These limits are not obvious in our tradition for reasons that

would also be welcome. But only to the extent that they celebrate

billiard balls and "large Poincare systems" would become leg-

should not be confused with a blind belief in progress or a fatal

endaiy fixtures there, commemorating the risk of a definition

submission to the argument of rationality. There is no need to

that requires that description and reason coincide, and pro-

define ourselves using terms that disqualify usan apparent

posing to other protagonists, like the makers of "artificial life,"

form of humilitybut that also transform into destiny what we

a contrast that may help them present their own practices. But

humbly recognize we are responsible for. If there is something

a scientific factish, regardless of the requirements and obliga-

in our tradition that positively singularizes us in the sense that

tions whose satisfaction it celebrates, and its representatives,

it must be taken into account in an ecology of practices, it is the

all those for whom this factish might make a difference, will not
make law, for in this Parliament it is the meaning that can be
given to such differences that counts, the way those differences
can be represented, become present for others. That is why what is
suitable for the Parliament of Things are not the requirements
and obligations of the "scientific laboratoiy" but of the "modern technical laboratoiy," where tests lead to creation rather
than proof, where proof is, depending on the circumstances,
required or irrelevant, where what is constructed is not a consequence but a novelty.
However, one question must be asked that highlights the
problem. The technical laboratoiy I'm referring to is modern
before it is technical. And even within the framework of an

invention of "politics." The limits are so obvious because the


condition that must be fulfilled by those who participate in the
Parliament of Things is a political condition, one that is usually espoused when condemning "corporatism." Every "citizen,"
every concerned group or practice, in the very act of defending
their interests, must be able to put the formulation of the interests in question to the test of general interest and acknowledge
their participation in the polity and its collective project. The
nature of the general interest, the collective project, the polity,
the citizen, all can be questioned. The condition does not apply
to the way in which the relationship between the group and the
individual will be taken into account but to the obligation to
share in constructing that relationship.

established democracy, in which the difference between those

Everything we today judge to be normal, a synonym of prog-

who have the power to impose obligations and the "public"

ress, has been invented through struggle, resulting in the inven-

consumers or users, who participate only in terms of their opin-

tion of new identities for the collective citizen, and transforming

ionwould not be ratified but actively opposed, the "Parliament

the modes of belonging, requirements, rights, and obligations

of Things," as I have characterized it, would also be modern,

that follow. But there is nothing neutral about this dynamic of

would embrace only "modern" representatives of conflicting

invention. It defines our "ecology" in a way that is political, that

interests. Such representatives must be capable of "promoting"

requires that we accept the test that distinguishes between condition

the constraints that, according to them, an innovative device,

and constraint. No one can impose a condition because everyone

disposition, or product must satisfy based on commonly intel-

is subject to the same conditioncollectively inventing what the

ligible criteria alone. They must also be prepared to participate


in the translation-betrayal-invention of what they represent.

world we all have in common will be. And that is why this test
cannot serve as a general obligation that defines the principle of

350

THE

POLITICS

OF T E C H N I C A L

INVENTIONS

an ecology of practices. Were that to happen, the conditions of


the politics we accept and whose constraints we demand would
change their meaning in the manner of the pharmakon. They

23

would become conditions we would apply to others. Unless we


are prepared to tolerate the fact that the categories of the "political" may not (yet) be obvious to them. And that would be our

The Gosmopolitical Question

curse.
Therefore, we may have to accept, as a new starting point,
that our political dynamic is in itself a "logical closure," as
understood by Tobie Nathan, not an ideal that can be generalized by right. The ecology of practices, if it has to avoid being
captured by the "great divide" between modern and nonmodern cultures, must certainly include the categories of the political, but it must not give them universal scope as part of "what
is human in man." 6 Those categories must be complicated, considered along "with" other categories that "fold" the question of
the decision differently. And among those other categories, we
find those "supernatures," those multiple worlds that seem destined to create obligations and conditions that are foreign to the
question of belongingeven when there is conflict within such
belongingto a polity.

If the question of politics were reduced to that of "nonmodern"


practices, that is, if only those practices were to "complicate"
the principles of an ecology of practices, the situation would be
extremely dangerous. For it would expose me to a renewed temp tation of "tolerance." This is what the multiplication of "musts"
that populate the preceding paragraph of this text translateswe
are the ones who "must." Consequently, I would have failed had
those imperatives referred to an abstract norm our categories
had to subject themselves to in the "name" of some duty toward
universality. Here, as elsewhere, the question of tolerance challenges and puts at risk an overly seductive proposition.
If the question of "others" does not have the meaning of a
normative imperative that must be obeyed, it is to the extent
that it may become a vector of requirements and obligations for
practices that can be represented by practitioners. And if practitioners are to complicate the model that relates the Parliament
of Things to the model of the modern technical laboratoiy, the
limits of that model must be made apparent from the viewpoint
of modern practices. This means that it must be understood that
the existence of "others" does not complicate our life but forces
us to recognize that complication.

351

THE

352

COSMOPOLITICAL

QUESTION

THE

COSMOPOLITICAL

QUESTION

353

The figure of the modern technical laboratoiy had as its

to incorporate in their creations questions as disparate as the

objective recognition of the difference between technical prac-

accessibility of resources, patents, workplace safety, and, lately,

tices and scientific practices. But it leaves open the question of

the problem of pollution and the need to recycle.

knowing whether all "technicians," even if we limit ourselves to


those who claim to be modern, are equally at home in the modern laboratoiy.
Let's examine a body of techniques invented by a lineage of
"constitutively modern" laboratoriesmetallurgy. This choice is
partly based on the fact that few would question the high degree
of success that has characterized this technical approach, one
that has integrated a number of scientific ingredients but has
also, on many occasions, paved the way for and guided scientific questions; and partly on the fact that metallurgy descends,
quite recognizably, from techniques that incorporate a constitutive reference to supernature. Modern metallurgy is certainly modern, and not "modernist," in that the purification of
supernature ingredients, although it has changed the nature of
metallurgy, has not led it to define itself in contradistinction to
its past, when the blacksmith was always something of a sorcerer. We could say that one technique has been substituted for
another by reusing certain components, retranslated, of course,
but recognizable in the new assemblage, and that the new technique does not exhibit in the form of symptoms the absence of
components it has not reused. Modern metallurgy has no need
to disqualify its ancestor, the blacksmith-sorcerer, and may
even be passionately interested in a practice that is no longer its
own. This is why it can retroactively be said that the art of metalworking has allowed itself to be "purified": it has not become
an "impossible profession." The practitioner of metallurgy is a
"natural" inhabitant of the modern technical laboratoiy, for the
translation-betrayal of the conditions of his practice constitutes
the veiy histoiy of the invention of that practice, which is robust
with respect to the redefinitions affecting it. The same description could also apply to chemical technicians, who have learned

My intention is not to review techniques but to create an


appetite for the distinctions among them that need to be constructed. Philippe Pignarre has proposed a suggestive cartography of what we call a drug, that is, what binds together those who
refer to it, from chemists and biologists in pharmaceutical laboratories to the doctor who prescribes it and the obedient patient
who purchases and consumes it.' The typical case is already more
complicated. Modern drugs can be presented as being "purified"
of the irrationality that condemned the pseudo-drugs of t h e past,
but the great singularity of present-day connections, their codified nature, the sophistication of the procedures of "socializing"
what is born as a "chemical molecule" and becomes, rarely, "a
drug," clearly illustrates that purification in this case no longer
has the same meaning as it does in metallurgy or chemistiy. It is
not only a case of production, but one of precarious production,
for it constantly alludes to what it must defend itself from and
is forced to incorporate unsettling tests (clinical trials using a
placebo) that it would willingly forgo (one day, rational pharmacology will be able to . . .). The technicians of "modern" drugs
today act on behalf of rationality by disqualifying others types of
drugs, those that cannot be separated from a practice of influence. In the Parliament of Things, these technicians will be
asked to negotiate with the representatives of patients' associations, who will no longer be impressed by this mode of presentation. The issue here is not to oppose modern drugs, presented
as the "only rational" drugs, to some form of "good medicine" to
which we must return. What is in question is the future of such
modern drugs, which face the challenge of political questions
from which they are currently protected by their qualification
as "rational." The modern technicians of drugs, including prescribing physicians, also have their place in the Parliament of

355

THE

COSMOPOLITICAL

QUESTION

THE COSMOPOLITICAL

QUESTION

353

Things, but the Utopia of this Parliament brings about the chal-

translation but translation that betrays. All of them would refuse

lenge of a modification of their "psychosocial" type.

to let themselves be represented by "citizens." Psychoanalysts'

But the question of drugs leads us indirectly to the question

lack of humor and the daunting nature of their claims would

of Freud's so-called impossible "profession": psychoanalysis.

then imply by default that, in their case as well, the "conditions"

As a technique of influence, psychoanalysis raises the ques-

of exchange have to be invented. How will the technicians of

tion of its modernism. Psychoanalysts singularly lack a sense of

psychotherapy be represented in the Parliament of Things?

humor about nonmodern "psychotherapies," and it appears that

Once a possibility exists, aside from the "we musts," to

their training does not prevent them (which is the least that can

treat "modern" and "nonmodern" symmetrically, the prob-

be said) from presenting themselves in ways that equate with

lem of the ecology of practices can finally become worthy of

the condemnation of eveiy other therapeutic practice. The mas-

the awe-inspiring word that gives its name to this series: cos-

ter words that cannot be disentangled from a dream of eradica-

mopolitics. For the word signals the path along which the ques-

tion, the somewhat cynical tolerance toward predictable alloys,

tion is to be constructed, that of the (re)invention of politics,

between "suggestion" and the gold of analysis, the analytic diag-

and the unknown that causes this path to bifurcate. Preserving

nosis of "resistance to analysis" all create a political problem.

the "political" character of the construction means that I do not

"Whereas modern metallurgy can accept that the "purification"

have a choice, that we do not have a choice. Eveiy proposition,

of its profession has in fact been a reinvention, whereas phar-

no matter how Utopian, if it is part of our tradition, draws from

maceutical technicians can claim that their definition of a drug,

the inventive resources associated with that tradition. But those

as partial as it may be, obligates them to reject the enormous

resources do not allow us to become "angels," authors of a Utopia

majority of candidate molecules, psychoanalysts seem to require

that is valid for eveiy inhabitant on Earth. They do not give us the

that we acknowledge their technique as "scientific," in the sense

ability to meet and recognize those who should be the coauthors

that it is said to satisfy what the human psyche requires and

of such a Utopia. The prefix "cosmo-" indicates the impossibility

would therefore have a relation to truth as opposed to fiction.


Let us assume that the analytic scene, like the divinatoiy
scene, be considered as an "influence laboratory" in the technical sense, that is, as foreign to the obligation that defines scientific laboratories as having to distinguish between the two
meanings of an artifact. Now. the problem is to determine if this
type of laboratory shouldn't be recognized as positively nonmodern. And in that case we could claim that, contraiy to metallurgy
and the art of manufacturing modern drugs, psychotherapy
cannot be subject to the "political" requirement that defines
a constitutively modern technique. Like ancestors or djinns,
the unconscious would not be compatible with the processes
of reinvention because reinvention does not only imply active

of appropriating or representing "what is human in man" and


should not be confused with what we call the universal. The universal is a question within the tradition that has invented it as a
requirement and also as a way of disqualifying those who do not
refer to it. The cosmos has nothing to do with this universal or
with the universe as an object of science. But neither should the
"cosmo" of cosmopolitical be confused with a speculative definition of the cosmos, capable of establishing a "cosmopolitics."
The prefix makes present, helps resonate, the unknown affecting our questions that our political tradition is at significant risk
of disqualifying. I would say, then, that as an ingredient of the
term "cosmopolitics," the cosmos corresponds to no condition,
establishes no requirement. It creates the question of p